CS Paper Radar

Daily listing

2026-10-16

790 papers first submitted this day (UTC). arXiv announces a day's submissions the next evening, so the newest day fills up over ~24 h.

Score = weighted mean ofCitations 20%Upvotes 10%Code stars 10%Venue 15%Code/data 10%Novelty 15%Momentum 15%Substance 5%how it works

AI & Machine Learning 248

Information Estimation with Discrete Diffusion

cs.LG cs.IT ·2026-10-16 ·Alberto Foresti, Giulio Franzese, Pietro Michiardi
top venuecode/datahighly cited
Abstract · Published as a conference paper at ICLR 2026 (Poster). ICLR Paper: https://proceedings.iclr.cc/paper_files/paper/2026/file/238747e153a84f50b…

Information-theoretic measures, such as Mutual Information (MI), play a crucial role in understanding non-linear relationships between random variables and are widely used across scientific disciplines. Yet, their use on real-world discrete data remains challenging. Existing methods typically rely on embedding discrete data into a continuous space and apply neural estimators originally designed for continuous distributions. This process requires careful engineering for both the embedding model and estimator architecture, but suffers from issues related to high data dimensionality. In this work, we introduce InfoSEDD, a discrete diffusion-based approach that bridges information-theoretic estimation and generative modeling such that they can be used to compute Kullback-Leibler divergences. Backed by Continuous Time Markov Chains theory principles, the design of InfoSEDD is lightweight and scalable and allows seamless integration with pretrained models. We showcase the versatility of our approach through applications on motif discovery in genetic promoter data, semantic-aware model selection in text summarization, and entropy estimation in Ising models. Finally, we construct consistency tests on real-world textual and genomics data. Our experiments demonstrate that InfoSEDD outperforms alternatives that rely on the "embedding trick". Our results position InfoSEDD as a robust and scalable tool for information-theoretic analysis of discrete data.

0.73

Do Knowledge and Reasoning Oriented Benchmarks Induce Stable Representation Signatures in LLMs?

cs.LG ·2026-10-16 ·Parsa Rahimi, Azim Dehghani Amirabad
publishedbenchmarkcode/datahot direction
Abstract · NeurIPS 2026 Interpretability as a Science Workshop

Internal signatures could support monitoring, adaptive computation, and mechanistic analysis, but benchmark prompts can be separable without any shared computational distinction. We test whether a prespecified knowledge/reasoning label defines a reusable internal signature in five contemporary 7B to 31B models. We treat the \textbf{task} as the unit of inference probes hold out complete tasks, labels are permuted across tasks, and the best layer and stream are reselected for every permutation. Raw probes recover an eight-versus-eight-subject MMLU partition in all five models (macro-F1 $0.78$--$0.81$), but prompt text predicts it nearly as well: under an identical probe configuration, the activation advantage is only $0.017$--$0.070$ macro-F1 and is not significant. At the prespecified $0.01$ level, lexical adjustment removes significance while the adjusted states still identify individual subjects, and the selected site changes across task panels and selection rules. Additionally, two released base-training trajectories show no common direction. These pooled layerwise measures recover the task label but identify no transferable knowledge/reasoning signature at the tested sensitivity, which cannot exclude moderate effects. The resulting protocol gives future internal monitors and intervention studies a sequence of tests to pass. Project page: \url{https://parsa-ra.github.io/lmbehave/}.

0.67

Neural Expansion: A Unified Mechanism for How Deep Neural Networks Generalize

cs.LG cs.AI ·2026-10-16 ·Chashi Mahiul Islam, Samuel Jacob Chacko, Mao Nishino, Canlin Zhang, Xiuwen Liu
top venuecode/data
Abstract · Accepted at NeurIPS 2026. Includes supplementary material. Code: https://github.com/meetchashi/neural_expansion_neurips_2026

Deep neural networks demonstrate remarkable generalization despite being highly over-parameterized, challenging traditional statistical learning theories. While several theoretical paradigms such as margin-based bounds, PAC-Bayesian analyses, and neural tangent kernels provide global explanations, they often lack mechanistic insight into how generalization emerges from training dynamics. We introduce Neural Expansion, a unified geometric framework that characterizes generalization through the expansion of directionally-selective ray coverage in the input space. Using Generalization Intervals (GIs), we quantify direction-specific regions around training samples where the model's predictions remain stable. Our extensive empirical studies across diverse architectures and datasets reveal that training leads to progressive expansion of ray coverage, particularly along directions with low Jacobian singular values. This mechanism helps explain a large fraction of correctly classified test inputs, adversarial examples, and even mislabeled or out-of-distribution samples through the geometry of ray coverage rather than instance-level memorization. By connecting input-space curvature to predictive behavior, our framework provides a mechanistic foundation that complements existing global theories of deep learning.

0.58

Towards Looped Models Done Right, Part I: Topology, Input Injection, Recurrent-State Design

cs.LG ·2026-10-16 ·Benhao Huang, Chufan Shi, Junlin Chen, Shicheng Wen, Zhengzhong Liu +2 more
code/datahighly cited
Abstract · Code and checkpoints: https://github.com/ifm-ai/xllm-loop

Looped language models reuse physical Transformer blocks across logical depth, but established architectures entangle recurrence placement, input injection, and recurrent state organization. We isolate these axes in models trained from scratch at matched parameter scale, logical depth, and token budgets, and then test whether the resulting design principles transfer to resident- and active-parameter-matched Mixture-of-Experts (MoE) models. The prelude-recurrent block-coda sandwich architecture consistently improves performance on context extraction and reasoning-intensive tasks. Input injection primarily enhances knowledge retrieval and context utilization, whereas random initialization and separate H/L recurrent states provide mixed or negative returns. Taken together, these results favor the Huginn-style design: at the 730M-parameter and 336B-token scale, the complete Huginn architecture outperforms Ouro across all ten benchmarks. The advantages of Huginn over Ouro transfer to MoE models. At the 8BA0.8B (32BA3B when unrolled) scale trained on 500B tokens, Huginn outperforms Ouro on nearly all benchmarks spanning knowledge, reasoning, and coding, while also exhibiting more balanced expert routing. Under matched training and inference FLOPs, the Huginn-style model uses 75% fewer resident parameters than a 112-layer feedforward MoE baseline. Despite its smaller physical parameters, it outperforms the feedforward baseline on DROP and GSM8K, matches it on MATH500, and remains competitive on the other benchmarks.

0.57

Equivalence testing and run-count design for stochastic machine-learning experiments

cs.LG stat.ML ·2026-10-16 ·Dmytro Knopov, Nadiya Shvai
code/dataunusual
Abstract · 31 pages, 4 figures, 4 tables. Code, simulation scripts and case-study data: https://github.com/DmytroKnopov/equivtest (release v0.2.0, arch…

Claims that two training methods perform equally well often rest on a non-significant difference between two or three runs per method. We develop a protocol, with its power theory, for certifying practical equivalence in per-run metrics such as the AUROC (area under the receiver operating characteristic curve). It rests on a two-level (seed x item) variance model separating between-run dispersion from item-level noise. Equivalence within a pre-specified margin is tested by two one-sided tests (TOST), equivalently by a confidence interval inside the margin; a two-level bootstrap covers composite statistics. A closed-form formula converts a margin and a target power into runs per method; its shifted-t refinement is calibrated against the realised two-level process. Simulations on a 270-cell grid and under non-Gaussian seed laws confirm the formula within its design-relevant range, from five runs per method. For four published runs of a joint energy-based model, the failure to certify equivalence was largely decided by design: with the variance calibrated to those runs, two per method certify with probability near three per cent, whereas certifying 0.01 AUROC at 80 per cent power needs twelve even under exact equality. A second case study on a thousand-seed CIFAR-10 zoo and the MultiBERTs checks the formula against real seed laws (within 0.05 in that range) and resolves comparisons that two or three runs leave undecided. In a two-stage design (pilot, then formula), our experiments suggest that an upper confidence bound on the pilot variance substantially reduces the undersizing seen with its point estimate.

0.57

Sample, Mix, or Verify? Allocating Test-Time Compute under Imperfect Verifiers

cs.LG cs.CL ·2026-10-16 ·Srinivasan Iyengar
Abstract · 27 pages (18 main + appendix), 14 figures, 11 tables. Code will be released

A common way to make language models and agents more accurate is to spend more compute at inference time: generate many candidate solutions and submit the one a verifier prefers. Success then requires that some candidate is correct and that the verifier picks it. The first grows predictably with the number of samples; the second does not, because real verifiers (unit tests, majority votes, LLM judges, reward models) accept wrong answers, reject right ones, and make correlated mistakes. With a fixed budget, a practitioner must decide how many samples to draw, which models to draw them from, and how much to spend on verification, and today these choices are made by rules of thumb. We introduce SPLIT, which makes them from a small labelled pilot. It rests on an exact formula for the probability of submitting a correct answer under accept/reject and top-score selection that accounts for varying task difficulty, blind spots shared across models, and correlated repeated verifier calls. SPLIT fits the formula by empirical Bayes and uses it to plan budgets on unseen tasks. Across seven benchmarks spanning math, science QA, code and software-engineering agents, 26 models, 175 agent systems and six kinds of real verifiers, all from public releases, SPLIT lands within 0.02-0.8 percentage points of the best allocation in hindsight, while every fixed rule we test loses at least 2.5 points in some setting. Regret depends only on how well SPLIT ranks configurations, and it stays near-optimal under compute, price and latency costs and online. The formula also explains when asking more LLM judges lowers precision. As a forecaster far beyond the pilot, SPLIT beats power-law fitting when the verifier caps success, and we report where it loses.

0.56

OliO: ODE-based Linear Transition Operator for Self-Supervised Time Series Forecasting

cs.AI ·2026-10-16 ·Kwangryeol Park, Seulki Lee
top venuecode/data
Abstract · Accepted at NeurIPS 2026

Self-supervised representation learning has advanced time series forecasting by capturing robust features from unlabeled data. However, existing contrastive and masking-based methods often struggle to explicitly model the underlying temporal flow, either by treating time lag as noise or by disrupting the inherent temporal dynamics essential for forecasting. To address these limitations, we propose OliO (**O**DE-based **Li**near Transition **O**perator), a plug-and-play self-supervised learning method that learns bidirectional temporal flows as a continuous and structurally consistent dynamical system. OliO introduces a transition operator derived from linear Ordinary Differential Equations (ODEs) to align latent representations across arbitrary time shifts. By imposing a strictly upper triangular constraint on the transition matrix, we mathematically ensure numerical stability and enforce a robust polynomial boundary that effectively captures long-term dependencies while preventing the exponential divergence typically found in unconstrained ODE-based models. This structural constraint induces a temporally coherent representation space that preserves the underlying flow essential for accurate forecasting. Extensive experiments across various backbone architectures and benchmarks demonstrate that OliO achieves significant performance gains, providing Mean Squared Error (MSE) reductions of up to 5.75% over the most competitive grid-searched SOTA baselines. The code is available at *https://github.com/eai-lab/OliO*

0.49

ROAR: Benchmarking LMM Agents under Real-Time Constraints in Multiplayer Action-Shooter Games

cs.MA ·2026-10-16 ·Dawei Wang, Di Zhao, Jianran Yu, Chengming Zhou, Marci Chi Ma +6 more
published
Abstract · NeurIPS 2026

Large multimodal models (LMMs) are increasingly deployed as agents that perceive, reason, and act in interactive multi-agent environments. Yet most existing benchmarks for LMM-based agents in multi-agent settings remain turn-based: the environment and other agents pause while an agent performs inference. Such protocols obscure a fundamental challenge of real-time constraints, where the world continues to evolve and delayed decisions may become outdated before execution. Motivated by this limitation, we introduce ROAR, a multiplayer action-shooter game environment for evaluating LMM-based agents in multi-agent settings under real-time constraints. ROAR covers cooperative, competitive, mixed cooperative-competitive, and basic single-agent scenarios. We further propose a Multi-Agent Asynchronous Gym interface that decouples environment evolution from agent reasoning by running the environment and each agent in separate threads, enabling the environment to advance at fixed time intervals without waiting for LMM inference and allowing agents to reason independently without blocking one another. ROAR also features convenient deployment, multimodal observations, and customizable levels. ROAR offers a practical testbed for advancing LMM-based agents beyond static, turn-based evaluation toward real-time multi-agent decision-making.

0.49

CocoRNA: Collective RNA Design with Cooperative Multi-agent Reinforcement Learning

cs.LG cs.MA ·2026-10-16 ·Tianmeng Hu, Biao Luo, Ke Li
publishedhot directionhighly cited
Abstract · ICML 2026

Designing RNA sequences that reliably fold into specific secondary structures is essential for understanding their biological functions but remains a challenging computational problem. We propose CocoRNA, a cooperative multi-agent reinforcement learning framework for RNA inverse design. CocoRNA simplifies the design task by decomposing it into smaller sub-problems, each solved collaboratively by multiple agents. This approach reduces the complexity of the problem and improves the exploration of design policies. During training, a centralized critic uses global structural information to guide the agents, enabling them to jointly optimize their design strategies. As a result, CocoRNA learns high-quality RNA design policies that generalize effectively to unseen structures without additional training. Experiments on the Rfam dataset demonstrate that CocoRNA substantially outperforms state-of-the-art methods in both success rate and design speed. Further experiments on other biological sequence design tasks highlight the effectiveness and broad potential of CocoRNA for complex design tasks.

0.49

Tree-Aware Posterior Distillation for Long-Horizon Agentic Reinforcement Learning

cs.LG cs.AIcs.CL ·2026-10-16 ·Heming Zou, Qi Wang, Yixiu Mao, Yun Qu, Yuyi Guo +5 more
community pick
Abstract

Reinforcement learning with verifiable rewards has become a scalable post-training paradigm for large language models. However, training agents over long-horizon interactions presents two challenges: using only terminal rewards makes turn-level credit assignment sparse and high-variance, while sequential mini-batch updates on a fixed rollout batch introduce policy staleness as the current policy drifts from the behavior policy. Existing approaches obtain denser credit via learned critics, process reward models, or on-policy distillation, but they rely on a value network, step-level annotations, or a stronger external teacher that increase training complexity. Meanwhile, importance sampling and clipping mitigate policy drift, but they do not specify a finite target that stops sequential updates from overshooting. In this work, we introduce TreeOPSD, a critic-free, tree-aware posterior self-distillation framework for long-horizon agentic reinforcement learning. Treating each interaction turn as a tree node, TreeOPSD branches alternative continuations from shared prefixes and recursively backs up terminal rewards into implicit turn-level credit. It then maps each credit to a closed-form, bounded teacher distribution parameterized by the likelihood ratio relative to the frozen behavior policy. Minimizing the divergence between this teacher and the student policy recovers the policy-gradient direction as the tangent at the behavior policy while bounding updates toward a finite target across sequential mini-batches. Empirically, TreeOPSD improves agentic capabilities across several competitive benchmarks and model scales, especially for long-horizon tasks, e.g., raising the average accuracy of Qwen3-30B-A3B-Thinking-2507 by 4.97 points over GRPO on the Deep Research setting.

0.48

JAXtari: High-Throughput and Easy-to-Modify Arcade Learning Environment

cs.LG ·2026-10-16 ·Quentin Delfosse, Raban Emunds, Paul Seitz, Sebastian Wette, Jannis Blüml +3 more
top venuebenchmarkcode/data
Abstract

The Arcade Learning Environment (ALE) has been a cornerstone of Deep Reinforcement Learning research, providing a standardized benchmark for evaluating general agents. However, its lack of native support for object-extraction and environment modification restricts research into evaluating whether agents learn robust causal models or rely on brittle visual heuristics. Furthermore, the ALE's dependence on CPU-based emulation creates a significant bottleneck in modern GPU-accelerated training pipelines, limiting both experimental scale and iteration frequency. In this work, we introduce JAXtari, an open-source, high-performance reimplementation of Atari 2600 environments. By leveraging massive GPU parallelization, JAXtari allows PPO agents to finish training on 100M environment steps on a single A100-GPU in under an hour. Beyond speed, it provides object-centric observations and a flexible modification system, allowing researchers to, e.g. develop neuro-symbolic methods or discover novel shortcuts and subsequently train and adapt their agents rapidly, sometimes in only minutes.

0.48

AM-Bench: A Unified Taxonomy and Evaluation Suite for Agentic Misalignment

cs.AI ·2026-10-16 ·Eric Zhang, Terry Jingchen Zhang, Chijioke Ugwuanyi, Jerick Shi, Bernhard Schölkopf +1 more
publishedbenchmarkcode/data
Abstract · Accepted as a poster at NeurIPS 2026 Evaluations and Datasets Track

As LLM-based agents are deployed with increasing autonomy in real-world environments, a critical failure mode emerges: agentic misalignment, where agents choose harmful actions over failure. However, existing safety benchmarks primarily evaluate whether agents comply with explicitly harmful instructions, leaving the conditions that cause agents to choose harmful actions poorly understood. To address this limitation, we propose AM-Bench, a unified taxonomy and 35 scenario evaluation benchmark spanning seven industries for agentic misalignment grounded in the fraud triangle, a well-established criminological framework explaining goal-directed misconduct. We adapt this framework to agentic environments and construct our evaluation suite and detection systems along three dimensions: perceived threat, perceived opportunity, and reasoning. We operationalize each dimension into structured subcategories rooted in prior theory: five threat categories derived from the instrumental convergence thesis, seven opportunity categories drawn from the Emergent Strategic Reasoning Risks (ESRR) taxonomy, and a novel four-question binary reasoning framework synthesized from prior literature. Our evaluation of 10 frontier models reveals a 62-percentage-point gap in misalignment across scenarios, with Gemini 3 Flash and Gemini 3.1 Pro exhibiting the highest levels of agentic misalignment while Claude Sonnet 4.6 displays near-zero rates across all conditions. Our reasoning framework further reveals that models may exhibit harmful behaviours without explicit situational awareness or explanation, complicating both mitigation and detection. AM-Bench scenarios and code are available at https://github.com/AI4Collaboration/Agentic-Misalignment-Bench

0.47

TorchTPU - A Hybrid Eager/Compiled Architecture for Native PyTorch Execution on TPU

cs.AI ·2026-10-16 ·Claudio Basile, Michael Voznesensky, Gunhyun Park, Ben Wilson, Parker Holloway +48 more
code/data
Abstract · 23 pages, 5 figures

TorchTPU is an open-source collaboration between Google and Meta that introduces a native and principled PyTorch backend for Google TPUs. It enables seamless out-of-the-box model execution and TPU migration while delivering industry-competitive inference and training performance through eager execution optimizations, torch.compile integration, and native hardware acceleration.

0.46

On the Selectivity of Generative Models in Structure-Based Drug Design

cs.LG cs.AI ·2026-10-16 ·Ella Miray Rajaonson, Jungyoon Lee, William Chau, Alan Aspuru-Guzik, Benjamin Sanchez-Lengeling +3 more
top venuebenchmarkcode/data
Abstract · Accepted as a Presentation at the Neural Information Processing Systems 2026

Designing small-molecule drugs that selectively bind to a target protein while avoiding unintended interactions with other proteins is essential for reducing side effects that contribute substantially to clinical attrition. Although structure-based drug design (SBDD) has advanced rapidly with deep learning, most generative models still optimize binding to a single target and disregard selectivity due to limited training data and unknown off-target interactions. Motivated by this gap, we introduce SELECTBENCH, a comprehensive benchmark assessing the selectivity of generative SBDD models. SELECTBENCH provides three curated evaluation tasks reflecting different real-world drug discovery scenarios: literature-based case studies, safety screening panels used in pharmaceutical development, and large-scale structure-based evaluation sets. We further introduce a suite of metrics to quantify selectivity and conduct a systematic evaluation of nine representative SBDD models, including methods explicitly trained for selectivity and widely-used inference-time guidance techniques. Our results show that even selectivity-trained models achieve only modest off-target discrimination and inference-time guidance provides limited gains, underscoring the need for method improvement. SELECTBENCH provides both a characterization of where current generative SBDD stands on selectivity and an open, extensible platform to accelerate progress. The benchmark is available at https://github.com/lovetaro/selectbench.

0.44

Deconstructing Multi-Task Active Learning: The Paradox of Gradient Conflict and Orthogonal Decomposition

cs.LG ·2026-10-16 ·Juho Jung, Seongjae Kang, Jinyoung Han
publishedpositioncode/data
Abstract · 10 pages, 5 figures, NeurIPS 2026

Existing multi-task active learning (MTL-AL) methods predominantly rely on linear scalarization of heterogeneous objectives such as uncertainty, diversity, and gradient conflict. We argue that this entangled formulation is fundamentally limited, as linear combination fails to capture Pareto-optimal trade-offs under conflicting objectives. We propose a shift from combination to deconstruction and introduce a principled framework that rethinks MTL-AL along three orthogonal dimensions. First, we identify the Conflict Paradox: although the optimization literature treats gradient conflict as interference to be suppressed, penalizing conflict when ranking candidates is measurably worse than not scoring it at all. We therefore decouple the two concerns: the acquisition function does not score conflict, and the resulting interference is resolved by gradient surgery during training. Second, we introduce Orthogonal Decomposition, separating acquisition into Informativeness and Exploration. This two-stage process, informativeness filtering followed by coverage maximization, prevents "diversity dilution" by excluding uninformative outliers before enforcing diversity. Third, we propose Cross-Task Mutual Information (CTMI), a covariance-based statistic of cross-task dependency computed from MC-dropout task entropies. Extensive experiments on COCO, Cityscapes, and NYUv2, including comparisons against recent MTL-AL and data-selection baselines, demonstrate consistent improvements over state-of-the-art methods, highlighting that structural rethinking of acquisition yields larger gains than incremental heuristic design. Code is available at https://github.com/Juho-Jung/multitask_active_learning_official.

0.44

Grounded Credit Assignment in Reinforcement Learning for Multi-Turn Search Agents

cs.LG ·2026-10-16 ·Qirui Liu, Hao Chen, Yi Wen, Weijie Shi, Yao Zhao +5 more
Abstract · 27 pages

Outcome-reward reinforcement learning such as GRPO gives every token of a search-agent rollout the same advantage, although the rollout exercises two capabilities: finding evidence and using it. We show that accuracy improves without grounding: Search-R1 raises accuracy on all seven QA benchmarks, yet on Bamboogle the evidence-supported share of correct answers falls from 84.0% to 42.9%. We attribute this to GRPO's credit assignment: rollouts that retrieve answer-bearing evidence are favoured in aggregate only in proportion to their accuracy gain over the rest, a grounding gap that answering from memory or misusing evidence shrinks or reverses. To address this, we propose Grounded Credit Assignment (GCA), which uses verifiable sub-outcomes (evidence retrieval and answer correctness) to correct search and answer credit separately: a wrong answer is blamed more than the search that found the evidence, and a correct answer that bypasses retrievable evidence is credited less. Never negative on evidence-finding search, the correction does not shrink with the grounding gap; capped at 10% of each token's advantage, it never overrides the outcome signal. Without extra models, rollouts, or training stages, GCA improves average EM over Search-R1 by 5.7 (Qwen2.5-3B) and 6.3 points (Qwen3-30B-A3B) and raises this share from 76.9% to 93.4%.

0.44

Batched Contextual Reinforcement II: Does RLVR Need Long Rollout Windows?

cs.LG cs.AIcs.CL ·2026-10-16 ·Bangji Yang, Hongbo Ma, Ruihan Guo, Jiajun Fan, Xi Zhu +1 more
Abstract · 22 pages, 3 figures, 2 tables

Long rollout windows make reinforcement learning with verifiable rewards (RLVR) expensive, yet simply shortening them can remove the correctness signal: when every single-problem rollout truncates before an answer, group-relative accuracy advantages vanish. As a technical continuation of Batched Contextual Reinforcement (BCR), we study how concurrent rewards relax this dependence on long windows. Concurrent Constraint Reinforcement (CCR) places three independently verifiable problems in one capped training completion. Correct answers completed before truncation retain credit, while a shared budget encourages shorter useful reasoning without an explicit length penalty. We connect this mechanism to conditional reward- and update-variance analyses and study 4B and 8B mathematical reasoning models. Against single-problem GRPO with a 32k window, CCR with an 8,000-token window reduces tokens per training step from 975,464 to 174,957 on 4B and from 793,832 to 165,788 on 8B, saving 82.1% and 79.1%, respectively. An 8B training trace shows increasing accuracy reward alongside shorter completions. Single-problem evaluations provide supporting evidence of retained reasoning quality. These results identify reward-preserving training structure as a practical route to lower per-step cost in short-window RLVR.

0.43

Beyond SAEs: Consistent Concept Discovery with Sparse Coding

cs.AI ·2026-10-16 ·Jinho Choi, Hyesu Lim, Jaegul Choo, Steffen Schneider
hot direction
Abstract

Sparse autoencoders (SAEs) explain foundation model representations by decomposing activations into sparse combinations of learned dictionary atoms. However, the learned dictionaries can vary across training runs, limiting the reliability and reproducibility of SAE-based analyses. We propose Consistent Sparse Dictionary Learning (ConSDL), which replaces the standard linear encoder with per-sample sparse coding. This is motivated by our synthetic analysis showing that coefficients from a single linear encoder pass are suboptimal compared with those obtained by per-sample optimization, suggesting that sparse coding is a reliable mechanism for dictionary learning. Across 16 synthetic settings, ConSDL achieves near-perfect dictionary and coefficient recovery. These gains translate to state-of-the-art cross-seed consistency across four vision and language settings, where ConSDL improves dictionary and coefficient consistency by 17.6% and 35.9%, respectively, over the average of six SAE baselines.

0.43

Dataset Inference for Data Provenance and Privacy Auditing in Tabular Foundation Models

cs.LG ·2026-10-16 ·Dariush Wahdany, Jesse Cresswell, Naiqing Guan, Atiyeh Ashari Ghomi, Franziska Boenisch +1 more
datasetcode/data
Abstract · Code at https://github.com/layer6ai-labs/tfm-di

Tabular foundation models (TFMs) are increasingly deployed through black-box APIs and trained on large collections of real-world tabular datasets. As this training paradigm becomes more widely adopted, private or proprietary datasets may be incorporated into pre-training corpora, yet there are currently no dedicated methods for determining whether a given tabular dataset was used to train a TFM. As a solution, we introduce the first dataset inference method for TFMs, aiming to infer whether a suspect dataset was part of a model's pre-training data. We systematically analyze a broad collection of candidate signals that can be observed from a black-box TFM via input manipulation and find that individual signals provide only weak evidence of training data membership. However, by combining multiple complementary signals, we can infer dataset membership for several state-of-the-art TFMs trained on real tabular data, achieving up to 0.99 ROCAUC dependent on the TFM. We then study factors that influence dataset identification, including pre-training data composition, model capacity, and the use of real vs. synthetic data. Our results show that our dataset inference method is a practical auditing tool for detecting privacy leakage and the use of proprietary datasets in TFMs.

0.42

Tool Recommendation for Multi-Step Agents: Finding Every Tool a Task Needs

cs.LG cs.IR ·2026-10-16 ·Sam Partee, River Stevens, Thierry Damiba, Valerie Fanelle
benchmark
Abstract · 11 pages, 3 figures, 5 tables

A multi-step agent task fails when a single tool it needs is missing, yet the tools an agent may need are too many to list: the roughly 8,200 tools on the platform we study have definitions that together exceed the context window of any model we evaluate. We present Tool Recommendation, a deployed system that gives an agent all of them, hundreds of times more than a typical MCP server exposes, through a fixed interface of three tools. It runs at the gateway between agent and tools, executing every call under the end user's own credentials and checking it against the user's consent for each service and the tools the agent may use. We evaluate it on both a public retrieval benchmark and billed end-to-end session cost. On ToolRet, Tool Recommendation places every tool a query needs in the top ten 57% more often than BM25, a lexical ranking that commercial tool search offers, and compares favorably with published systems up to nine times its size. In paired multi-step workflows, its first recommendation covered every required tool in 93% of Codex sessions, and against a fixed list of 100 tools it cut input tokens by 17% with Codex and 40% with Claude Code and the total bill in all 14 runs, by up to 40%, with no change in task completion.

0.42

ThetreStageAI: Multi-View Retrieval-Augmented Generation for Structured Bengali Theatre Production Planning

cs.LG ·2026-10-16 ·Md. Mehedi Hasan Naeem, Mst. Kamrunnahar Ruma, Shakila Sultana, Nafiza Anjum, Nasreen Jannatun Nemu +1 more
publishedcode/data
Abstract · 6 pages, 6 figures. Accepted at the 12th IEEE International Conference on Sustainable Technologies and Engineering (IEEE iCOSTE 2026), Cox's…

Staging a theatre performance requires coordinating dialogue, actor movement, stage directions, lighting focus, cue timing, and equipment constraints within the same production plan. This work presents ThetreStageAI, a retrieval augmented framework for structured Bengali theatre production planning. Rather than retrieving from one mixed corpus, the system maintains separate scene, actor-blocking, and lighting views. Bengali text is embedded with BGE-M3 and indexed in view-specific Qdrant collections. Retrieved references are combined with a production brief and passed to Gemini, which generates schema-constrained JSON containing characters, scenes, dialogue, stage directions, blocking, lighting, and optional sound cues. Structural and semantic checks are applied before acceptance, with at most one constrained repair attempt. We also construct a human-guided, LLM-assisted synthetic dataset of 500 short Bengali theatre records that expands to 1,500 retrieval documents. Dataset diagnostics confirm consistent identifiers and valid lighting fields while also revealing substantial dialogue repetition. A paired camera-ready evaluation compares single-combined and full multi-view retrieval under matched settings. The source code, dataset, and evaluation materials are publicly available.

0.41

How the elastic axis folds the transonic flutter boundary: shock energy transfer and conformally calibrated active learning

cs.LG physics.flu-dyn ·2026-10-16 ·Ashray Saxena, Sharad Shrivastava
publishedunusual
Abstract · 28 pages, 15 figures, 8 tables. Accepted for publication in Transport Phenomena (Special Edition: Tribute to John Charles Slattery). Code an…

Isogai's case A, a pitch-plunge aerofoil whose elastic axis lies half a chord ahead of the leading edge, has a folded transonic flutter boundary in inviscid flow, with several flutter speeds at one Mach number. We map this boundary and its elastic-axis family with coupled Euler simulations, a Gaussian-process surrogate of the damping trained by batch active learning, and distribution-free split conformal prediction, which bounds the marginal error rate of stability calls. Three failure modes of damping labels are invisible in the residuals, and the half-degree mean incidence left by an impulsive start at $1^\circ$ freestream incidence raises the flutter speed at the dip by 2.8%, more than any numerical setting. With 80 simulations the learned case A boundary labels 63 of 64 independent runs correctly, and at most 1.8% of calibrated calls on held-out runs are wrong. Over Mach number, elastic axis and flutter speed index, 193 simulations show that, for the case A structure about the axis, the inviscid fold exists only for axes more than about 0.4 chord ahead of the leading edge, on two meshes. Chordwise energy budgets, closed three ways, explain both branches. On the lower branch the falling reduced frequency, at unchanged shock phase, raises the shock input and lowers the upstream damping similarly. On the upper branch the shock is pinned at the trailing edge; moving the axis aft brings the flutter-mode pivot towards the section, so the upstream damping falls faster than the residual shock input and the fold closes.

0.41

Process Reward Modeling with Reliability- and Conservatism-Aware Bayesian Belief

cs.LG ·2026-10-16 ·Zhongjian Qiao, Zhixiang Zhou, Jiafei Lyu, Peisong Wang, Qiaosheng Zhang +2 more
theory
Abstract

Process Reward Models (PRMs) play a central role in test-time scaling for large language model (LLM) reasoning. However, PRMs may provide inaccurate process rewards for test-time scaling, which we identify as arising from two sources of uncertainty: (1) \textit{PRM uncertainty} that arises from insufficient supervision data and imperfect PRM training; and (2) \textit{policy uncertainty} that results from the mismatch between the policy to collect supervision data and the policy for test-time scaling, leading to systematic reward overestimation. Previous studies fail to jointly consider both uncertainties. To bridge this gap, we propose BayesianPRM, which treats the process reward as a random variable within a reward hypothesis space and maintains a Bayesian belief over it. Since exact Bayesian inference over the continuous hypothesis space is generally intractable, we discretize the reward hypothesis space with a reward ensemble and introduce two complementary belief updates, each targeting one source of uncertainty: (1) to address PRM uncertainty, we introduce \textit{reliability-aware belief updating} that measures how well each ensemble member explains the supervision data and updates the belief with an amortized belief network optimized by the evidence lower bound (ELBO); and (2) to tackle policy uncertainty, we propose \textit{conservatism-aware belief calibration} which further reweights the belief toward lower rewards to suppress overestimation. These two updates yield a reliability- and conservatism-aware Bayesian belief, whose expectation determines the process reward. Experiments on multimodal reasoning benchmarks indicate that BayesianPRM consistently yields more accurate process reward estimates while enhancing test-time scaling performance.

0.40

Nearly Linear Private Learning of Littlestone Classes

cs.LG ·2026-10-16 ·Shay Moran, Chao Yan
theory
Abstract

Differentially private learning aims to learn from data while protecting individual training examples. For binary classification, finite Littlestone dimension characterizes approximate private PAC learnability (Alon et al., JACM 2022). Understanding how this dimension controls private sample complexity remains a central question. For classes of Littlestone dimension $d$, the first general upper bound was doubly exponential in $d$ (Bun, Livni, and Moran, FOCS 2020). Subsequent work improved this to $\widetilde O(d^6)$ (Ghazi et al., STOC 2021) and then to $\widetilde O(d^5)$ (Lyu, STOC 2026). We prove a nearly linear upper bound of $O(d\log d)$. Our main algorithmic contribution is a structural exponential mechanism for data-dependent candidate sets equipped with a rank satisfying suitable stability conditions. For natural Littlestone classes such as subspace indicators, this rank has a simple geometric interpretation: it is the codimension. This yields a private PAC learner for subspaces of dimension at most $d$ over any field, with tight linear dependence on $d$ and no dependence on the ambient dimension, improving the quadratic dependence of Kaplan et al. (STOC 2025). Our second contribution is combinatorial: drawing on model-theoretic ideas, we construct a suitable rank for general Littlestone classes and establish the counting bounds required by the mechanism.

0.40

Towards Hyperparameter Transfer for Differentially Private Optimization

cs.LG cs.CRstat.ML ·2026-10-16 ·Tianze Wang, Zhiqi Bu, Linjun Zhang
top venue
Abstract · Accepted at NeurIPS 2026

Scaling laws and hyperparameter tuning remain central challenges in training large-scale models, particularly under differential privacy (DP), where optimization is further complicated by noise injection and per-sample gradient clipping. While Maximal Update Parametrization ($μ$P) allows for zero-shot hyperparameter transfer in standard training, we show that it fails under DP constraints due to the distinct spectral properties of high-dimensional noise versus low-rank gradients. In this work, we introduce DP-$μ$P, a theoretically grounded framework that extends $μ$P to the privacy-preserving setting. By analyzing the interaction between gradient clipping, noise injection, and feature learning, we derive a Signal-to-Noise Ratio (SNR) condition that governs layer-wise learning rate scaling. Combining this with our proposed spectral scaling formulation, we provide specific scaling rules for DP-SGD, DP-Adam, and DP-Muon optimizers. Empirically, we demonstrate that DP-$μ$P enables effective zero-shot transfer of learning rates across model widths on diverse architectures (MLP, ViT, GPT-2), matching or exceeding the performance of individually tuned baselines while significantly reducing computational costs.

0.40

Constructive Universal Approximation by Infinite-Dimensional Neural Networks via Gaussian-Weighted Ridgelet Transform

cs.LG math.FAstat.ML ·2026-10-16 ·Sho Sonoda, Luca Galimberti, Yuanchao Xu, Isao Ishikawa, Masahiro Ikeda
hot direction
Abstract

Neural networks whose inputs are functions, as in operator learning, are usually shown to be universal by existence arguments, either by reduction to finite-dimensional networks or by Stone--Weierstrass-type density; these prove that approximants exist but say little about how the parameters depend on the target. We take a constructive route. We extend the ridgelet transform, an integral transform related to the Fourier, Radon, and wavelet transforms, to continuous maps on a separable infinite-dimensional Hilbert space, replacing the Lebesgue measure that is unavailable in infinite dimensions by Gaussian weights with a homogeneous mixture on the direction space. The resulting transform has a Plancherel identity and reconstruction formulas incorporating the frame operator, the correction induced by the weights, and it assigns to each target an explicit coefficient measure, one for every admissible analysis filter. From these representations we obtain uniform approximation on compact sets for all continuous non-polynomial activations of polynomial growth, including ReLU, and, by Monte Carlo sampling of the coefficient, finite-width networks with a dimension-free $O(N^{-1/2})$ error in the width $N$ for globally Lipschitz activations and coefficients with finite variation and second moments. Explicit examples include Gaussian targets and neural-operator layers. A companion Lean~4 formalization provides machine-checked proofs, with its correspondence to the manuscript documented at https://shosonoda.github.io/lean-operator-ridgelet/ .

0.39

Inside the Loop: A Mechanistic Study of Weight-Tied Transformers on Depth-Bound Algorithmic Tasks

cs.LG cs.CL ·2026-10-16 ·Magnus Sesodia, Juil Sock, Philip Torr, Christian Schroeder de Witt
top venuecode/data
Abstract · Accepted to NeurIPS 2026. 17 pages, 12 figures

Solving algorithmic tasks with many sequential steps typically requires scaling up transformer depth or chain-of-thought length, incurring unfavorable parameter and inference-compute costs, respectively. Looped transformers, which reuse one block of layers across many iterations, are poised to address this, since they scale effective depth without adding parameters or tokens. Yet, it remains unclear what each iteration is actually computing, or whether the loop merely simulates a deeper untied stack. Here, we study directed permutation traversal, in which a model follows $H$ hops from a starting node and reports the node it reaches, so every intermediate hop has a known ground truth. We find that a minimum effective depth is necessary to solve the task, beyond which the degree of weight tying becomes the determining factor. For instance, a looped transformer with a four-layer block iterated eight times solves the task in every seed, while an identically-trained untied transformer of the same effective depth never does, succeeding only with a lower learning rate or supervision of its intermediate layers. Probing and intervening on the residual stream, we find that the looped model advances through the intermediate nodes gradually, forming a staircase that adapts to its iteration budget, whereas the untied model, at that lower rate, exposes them only in its final layers. Taken together, our results recast looped transformers as iterative program learners, which allocate their iterations to successive steps of the algorithm and solve tasks that identically-trained untied models of the same effective depth do not. Code is available at https://github.com/magnus-sesodia/insidetheloop.

0.39

Training-Free Prompt Selection for Efficient Reinforcement Learning With Partial Rollouts

cs.AI cs.LG ·2026-10-16 ·Lei Jiang, Honghui Bao, Yue Zhou, Zhengyao Gu, Langzhou He +1 more
benchmark
Abstract

Reinforcement learning with verifiable rewards (RLVR) has emerged as a prominent paradigm for training reasoning models, but its computational cost makes efficient data selection increasingly important. Since training examples contribute unequally to learning, selecting a small, informative subset can offer a more efficient alternative to using the entire data pool. The training-free selection method estimates prompt value with hidden states from complete greedy rollouts. However, in our datasets, $26$--$32\%$ of these rollouts fail to terminate within the generation budget. We introduce LIFT, a training-free data selector that identifies impactful prompts using only partial reasoning rollouts. At its core, LIFT uses Partial Rollout Context Lift (PRCL), a score inspired by pointwise $\mathcal{V}$-information that accumulates token-level log-likelihood ratios between a question-conditioned context and a null context with the question removed. This enables prompt scoring from a short rollout prefix rather than a complete trajectory. Across Qwen2.5-Math-1.5B and DeepSeek-R1-Distill-1.5B, prefixes retaining at least 90\% and 95\% of the full-rollout PRCL score, respectively, require only 26\% and 4.7\% of the tokens generated by full rollouts. Compared with SHIFT, LIFT reduces selection cost by 3.7--9.1$\times$ while achieving competitive downstream RLVR performance, with the highest six-benchmark average in three of six settings.

0.39

On Reparameterizing the Score Function

cs.LG stat.ML ·2026-10-16 ·Mohammad Hossein Amani, Leonardo Bocchieri, Maxime Peyrard, Clément Hongler, Robert West
top venue
Abstract · Accepted at NeurIPS 2026. 23 pages

Policy gradients for autoregressive sequence models differentiate each token's log-probability while holding the conditioning history fixed. This discards information about how earlier decisions shape later conditional distributions, creating a credit-assignment bottleneck when rewards are sparse. We introduce History-Reparameterized (HR) policy gradients, which replace the frozen prefix in the score function with a differentiable reparameterized or relaxed history, allowing gradients of downstream log-probabilities to propagate backward through the generated trajectory. The resulting history-pathwise term is generally biased when added directly, so we derive Optimal HR (OHR), a centered control-variate estimator with a closed-form coefficient that minimizes the trace of the gradient variance at the population level. The variance reduction is governed by temporal structure in the policy dynamics: random or weakly structured policies provide little benefit, while policies with learned sequential dependencies yield informative history-pathwise corrections. For discrete tokens, the choice of relaxation is crucial: embedding-level relaxation yields substantial variance reduction that grows with horizon, whereas straight-through Gumbel-Softmax yields essentially none in our diagnostics. We validate the method on a Gaussian linear Markov model, a sparse-reward T-maze with a recurrent policy, and an autoregressive Transformer model for small Traveling Salesman Problem (TSP) instances.

0.39

GOLD: Geometric Optimized Latent Diffusion for Structure-Aware RNA Inverse Folding

cs.AI ·2026-10-16 ·Qi Si, Xuyang Liu, Penglei Wang, Shuo Su, Limei Han +4 more
top venuecode/data
Abstract · Accept as a Spotlight Paper in NeurIPS 2026

Designing RNA sequences that fold into specific 3D structures is a fundamental challenge in therapeutics. Current RNA inverse folding methods typically condition on 3D backbone coordinates but the generation is limited to 1D discrete sequences. This ignores the continuous side-chain orientations, which are essential for overall structural stability. To address this issue, we explicitly introduce side-chain geometry into the RNA generative process. Unlike co-design frameworks that diffuse across separate raw spaces, we propose Geometric Optimized Latent Diffusion (\textbf{GOLD}), which project sequence and structure into a unified latent manifold. GOLD employs a novel GeoVAE to elegantly integrate discrete RNA sequences with continuous side-chain angles. By mapping these modalities into a single, smooth latent manifold, GOLD naturally facilitates a synergistic co-diffusion process. To enhance structural validity, we further introduce a training-free, gradient-guided sampling mechanism. By decoding the generated sequence probabilities and side-chain geometries into structures during the sampling process, we compute analytical gradients from physical constraints. This dynamic feedback actively guides the generated sequence to a physically stable state. Extensive experiments validate the superiority of GOLD, achieving state-of-the-art performance across sequence recovery, secondary structure validity, and 3D foldability metrics. Our code and data are publicly available at https://github.com/darkflash03/GOLD.

0.38

Large language model and reinforcement learning based autonomous construction and progress updates

cs.AI ·2026-10-16 ·Yifei Xiao, Huasheng Lin, T. Y. Yang
top venue
Abstract

Due to the rapid growth of the global population, traditional construction methods are gradually unable to meet the current housing demand. The advancement of machine learning algorithms has encouraged researchers to explore their applications in the construction industry, leading to more intelligent and efficient building processes. In particular, the rise of large language models in recent years has provided more possibilities for the transformation from traditional construction to smart construction. In this study, a large language model and reinforcement learning based autonomous construction method is proposed. An enhanced soft actor-critic reinforcement learning algorithm is developed to plan actions for a mobile crane according to construction tasks. Robot kinematics and control algorithms are used to ensure that the actions generated by the reinforcement learning algorithm can be executed by the crane. In addition, a customized embedding model is developed to extract information from the construction environment according to user interests. Then, a large language model, ChatGPT 5, is used to update construction progress, provide safety information to control system, and generate construction progress reports. The feasibility of proposed method is verified through several full-scall retaining wall construction experiments.

0.38

OfficeTown: Generative Organizations for Evolving Agent Intelligence

cs.AI ·2026-10-16 ·Hongchen Wei, Daohong Ying, Zhi Hong, Aidan Wu, Bei Liu +8 more
community pick
Abstract

AI agents increasingly use files and software tools to produce professional deliverables. Such work draws on materials, decisions, and feedback accumulated across assignments. We introduce OfficeTown, a generative simulation framework for professional organizations in which tasks and their context arise from continuing organizational work. From a natural-language description of an organization's purpose and activities, OfficeTown initializes an organization whose members coordinate work, delegate bounded assignments to assistant agents, and review their deliverables. Artifacts, decisions, and unresolved matters persist and shape subsequent work, while responsibilities and permissions determine what each member can see and do. We curate work requests, their context, and review feedback into executable resources for evaluation and learning. We construct OfficeTown Bench with 406 tasks and 4,836 evaluation criteria from 23 simulated organizations across eight professional domains, incorporating domain-expert feedback. Frontier-model evaluations reveal substantial remaining headroom. Supervised fine-tuning and on-policy distillation on OfficeTown data substantially improve Qwen3.5-9B on external professional-work benchmarks, including GDPval. Supplementary evaluations show that the principal--assistant harness achieves higher scores than the single- and multi-agent baselines tested. These results highlight the value of organizational simulation in generating work-grounded data and environments for the continued development of generalist agents.

0.38

The Weight of Teamwork: Does Communication Structure Help LLM Scientists Search for Scientific Laws?

cs.MA cs.AIcs.LG ·2026-10-16 ·Jianing Shi, Zaibin Zhang, Shijun Lei, Zhenfei Yin
publishedunusual
Abstract · Accepted at the NeurIPS 2026 Workshop on Foundations of Agentic Systems Theory (FAST), non-archival. 7 pages main text, 33 pages total with …

Team performance is often reported as one delivered score, even though that score is produced in two stages: a system generates candidate answers and a selection rule returns one of them. We use symbolic regression as an observable scientific-search setting and compare fixed-budget LLM teams that differ in peer visibility and update timing, evaluating both their frozen candidate archives and the expressions selected from those archives. Distributing the call budget across more independent trajectories produced the largest measured gain, improving mean held-out fitting score Q by 0.70 at the widest tested allocation. Under a common archive-wide selector, the implemented Ring policy reduced mean Q by 0.33 relative to Independent, while Star and Complete remained unresolved. Holding the generated archives fixed, changing the delivery rule changed the measured architecture gaps, including the sign of the Complete point estimate. These results support reporting candidate generation and final selection as separate endpoints. They do not identify a communication mechanism or a universally preferred topology; rather, they show that an architecture ranking is not identified by communication structure alone when the rule that returns the final answer is left unspecified.

0.38

Next Token or Vector? Learning How Far to Decode

cs.LG cs.AIcs.ET ·2026-10-16 ·Sai Sreenivas Chintha, Chidera Biringa, Rishab Sharma, Anusha Agrawal, Prachi Badarayani +4 more
Abstract

Standard autoregressive decoding advances one token at a time, even when several upcoming tokens can be predicted together. We propose CADENCE (Context-Adaptive DEcoding), a generation method that learns both what to draft and how far to draft. Gaussian prediction heads represent future spans as continuous latent vectors, and a shared decoder converts the selected vector into token predictions in parallel. A learned router selects the prediction length from the model's own hidden states and predicted Gaussian parameters, adapting generation granularity to the current context with single-token generation as a fallback. The backbone verifies each proposed continuation and retains the prefix that matches its greedy predictions. Experiments with Llama-3.1-8B-Instruct on Alpaca, GSM8K, and HumanEval show that router-selected positions yield higher draft acceptance than applying the corresponding prediction length everywhere. CADENCE achieves up to 1.3x higher generation throughput than standard greedy decoding.

0.38

Probing Ambiguity: Adversarial Intent Resolution for Robust Text-to-SQL

cs.AI cs.CL ·2026-10-16 ·Youbing Yin, Xujun Peng, Muhammad S. Uddin, Erin Babinsky, Daben Liu +1 more
top venuebenchmark
Abstract

Text-to-SQL systems typically assume a one-to-one mapping between natural-language queries and correct SQL translations. In practice, however, many queries are inherently ambiguous, admitting multiple valid interpretations that yield distinct outputs. We introduce Adversarial Intent Resolution (AIR), a multi-agent framework that actively probes and resolves such ambiguity without any task-specific training. AIR first constructs competing interpretations, verifies through SQL execution that they produce genuinely different results, and generates targeted clarification questions grounded in the observed data differences. Evaluating on leading benchmarks for ambiguous questions, AIR demonstrates strong performance, efficiency, and robust cross-benchmark transferability. On the AMBROSIA benchmark under schema-only context, a realistic production setting where exposing the full database is infeasible, AIR outperforms IntentRL, the state-of-the-art reinforcement learning model, across all metrics (e.g., F1 0.679 vs. 0.618). Crucially, AIR exhibits only ~7% performance degradation without the full database, compared to ~20% for IntentRL. On the AmbiQT benchmark, where ambiguity arises from schema design rather than linguistic structure, AIR achieves higher F1 under both context settings without any benchmark-specific adaptation.

0.37

Reasoning Models Can Control Their Chains of Thought When They Tell Themselves To

cs.AI cs.CLcs.LG ·2026-10-16 ·Uriel Dolev, Idan Schwartz, Yoav Goldberg
positioncode/data
Abstract · 26 pages, 19 figures, 18 tables. Code and data: https://github.com/udolev/Self-Instructed-CoT-Controllability

Chain-of-thought (CoT) controllability, the ability of reasoning models to control properties of their CoT, is consequential for both their usefulness and reliable oversight. Recent work evaluates this capability by measuring whether models follow CoT constraints given in the user prompt, and reports very low controllability across models. We argue that this low controllability may not reflect an inherent inability of reasoning models to control their CoT, but instead result from naively specifying constraints in the user prompt. To test this, we introduce Self-Instructed CoT Controllability: a complementary evaluation that places the constraint as a self-instruction inside the model's CoT and measures its ability to follow it. We show that, under self-instruction, models achieve significantly higher controllability than with standard user instructions. Even a generic reminder to follow the user's constraint substantially improves controllability. Moreover, effective self-instructions can be bootstrapped into the reasoning through a simple instruction in the user prompt, enabling improved control even without access to the reasoning channel. Finally, using self-instruction, we demonstrate in a realistic monitoring setup that CoT controllability can be leveraged for monitor evasion, posing a risk to CoT monitorability.

0.37

Reducing double-dipping in single-cell topic models with three-way count splitting

cs.LG q-bio.GN ·2026-10-16 ·Zheng Qu
code/dataunusual
Abstract · 16 pages, 4 figures. Code: https://github.com/quz2004/topic-model-count-splitting

Topic models learn gene programs from single-cell RNA sequencing (scRNA-seq) data, and the same counts are then used to test which genes and donors are associated with those programs. Count splitting is the standard remedy for this reuse, but it has not been tested when an encoder reads each cell's counts. On a 104,400-cell lupus cohort with 300 permuted null genes per run, naive tests rejected 36% of null gene-topic pairs at a nominal 5% level, and the usual two-way split rejected 38%, because the encoder sees the tested gene's own held-out count. Encoding topic usage on a separate third of the counts, with beta-binomial thinning, lowered this to 7.7% (6.7% for genes outside the model); for two other models, LDVAE and scETM, the rate was 7.0-7.3% (naive 51-59%). A label-free interferon program shifts in held-out counts in the lupus cohort and in a pre-specified COVID-19 cohort, and donor-level topic usage predicts disease (AUROC 0.96) beyond cell-type proportions (0.86 and 0.58 in the two cohorts). Calibration and prediction did not depend on the model. We do not claim better clustering, prediction or topic stability. Code, figures and result tables are public (MIT license; DOI 10.5281/zenodo.23307193).

0.37

Parametric Resonance in Primal-Dual Learning for Slowly Non-Stationary Constrained MDPs

cs.LG stat.MLeess.SYmath.OC ·2026-10-16 ·Maxime Baelde
code/dataunusual
Abstract · 33 pages, 3 figures. Code: https://github.com/mbaelde/dualdynamics-code

Primal-dual methods for constrained Markov decision processes are analyzed under a two-timescale assumption, with non-stationarity entering through two channels. In the additive channel, where the constraint level oscillates, the dual oscillation saturates at a floor set by the moving optimum and is amplified linearly by the timescale ratio. In the parametric channel, where the constraint sensitivity oscillates, a coherent modulation carries no resonant forcing at any depth: the dual update's stiffness modulation cancels the primal update's damping modulation. Instability instead comes from a mismatch between the constraint estimate used by the primal update and the measurement used by the dual update, producing a parametric resonance at twice the pair's natural frequency. The tolerated mismatch grows with the square root of the forcing bandwidth and collapses with the square root of the dual-to-primal step-size ratio. These results hold on the tangent linear-quadratic reduction around an active saddle, for one constraint and for $K$ constraints under a common modulation and an isotropy condition. Driving both updates from the same estimate removes the resonant forcing but introduces a stationary offset equal to the estimator's bias. In a Lagrangian soft actor-critic agent, the multiplier locks onto the sub-harmonic at the timescale ratios its identified linear loop predicts and a coherent cost critic suppresses the lock-in, while a proportional (PID) term, predicted to leave the lock-in in place, suppresses most of it through the projection of the multiplier.

0.37

Between Spoilers and Guesswork: Avoiding Over-Specification in Coding Benchmarks

cs.AI cs.SE ·2026-10-16 ·Yuanzhe Liu, Bastian Lipka, Avraham Shinnar, Jatin Ganhotra, Yao Ma +1 more
benchmarkcode/data
Abstract · Codes available at https://github.com/IBM/swe-hints

Benchmarks drive AI progress, but are hampered by task under-specification and over-specification. Task under-specification causes guesswork, where solutions have to make things up, or worse, resort to contamination or reward hacking. Task over-specification causes spoilers, where solutions can bypass some of the harder steps the benchmark was supposed to measure. Since future-looking benchmarks should be difficult, over-specification reduces a benchmark's shelflife of community adoption. Repository-level coding benchmarks have helped the field progress but experience unnecessary churn due to both over- and under-specification. For instance, SWE-bench Pro, FEA-bench, Doc2Feat-bench, and Multi-SWE-bench all tackle under-specification by augmenting task statements with hints derived from ground truth. We show that leaking ground-truth information into the task statement causes over-specification in some of these benchmarks. We experiment with various specification levels, trying to thread the needle between spoilers and guesswork. Based on these results, we recommend techniques for deriving parsimonious hints. We will release our hints so the community can obtain more reliable signals from popular benchmarks.

0.37

Certifiable inversion of resampling-induced spectral distortion in high-dimensional particle filters

cs.LG ·2026-10-16 ·Deming Song, Chi-Kin Lam, Kefeng Li, Hui Yu, Xilin Li +1 more
theorycode/data
Abstract · This is a preprint of an article submitted to Inverse Problems. The implementation and experiment scripts are available at https://github.co…

This paper formulates and solves the inverse problem of recovering the pre-resampling spectrum of a particle filter's ensemble covariance from the importance weights and the observed spectrum alone. Resampling controls weight degeneracy, but in high dimensions it also distorts that spectrum: the distortion is a free multiplicative convolution with a law computable from the weights, so the repair is a deconvolution. We first extend the forward spectral law to random importance weights by conditioning on the weights together with the realized offspring counts rather than on the weights alone, which makes the distortion law deterministic at the observed data and gives the zero-atom mass of the limiting law a direct reading as an effective-dimension proxy for the second-order information the filter has lost. Second, a regularized free deconvolution inverts the distortion with an explicit error budget at a fixed resolution scale, and a computable forward residual certifies when the model behind it is violated. Injectivity makes the inverse identifiable but not stable, so the budget rests on a resolution-restricted condition number whose positivity we prove, and is vacuous below that resolution rather than concealed in a hypothesis. Third, the repaired spectrum enters only as the preconditioner of a Metropolis-adjusted rejuvenation move, so spectral error changes the mixing constants and not the invariant target, and the full cycle obeys a geometrically summable error recursion whose abstention term is governed by when abstentions fall rather than how many occur. Pre-registered simulations confirm the forward law, an order-of-magnitude reduction of spectral error, and the abstention law. The repair recovers eigenvalue laws, not eigenvectors.

0.37

Waiting Buys Exactly Two Moments: Endogenous Identification in Prediction-Driven Allocation

cs.LG ·2026-10-16 ·Rama AlHamidi, Rasul Khanbayov, Erchin Serpedin, Hasan Kurban
code/data
Abstract · 24 pages, 2 figures

A planner with a fixed budget of preventive help must decide whom to help and when. Waiting reveals risk but removes intervention opportunities as higher-risk individuals leave first. We characterize this tradeoff under repeated noisy signals and risk-dependent attrition: in the canonical model, a length-$τ$ observation prefix identifies exactly the latent risk moments $m_j=\mathbb E[p^j]$ for $j=0,\ldots,2τ-1$, buying two new moments per additional period. These identify average next-period attrition among survivors but need not identify allocation welfare or attrition among those prioritized for treatment. Our observationally equivalent constructions establish unavoidable timing regret at any sample size for a fixed observation length, while exact certificates determine whether all compatible populations share a unique optimal date. Matched experiments separate the effects of estimation, bootstrap averaging, and regret weighting. Longer observation often resolves timing ambiguity, but the optimal date may already have passed. In our experiments, bootstrap averaging and early stopping improve decisions, while regret weighting adds no improvement in pooled mean squared regret: better statistical fit need not yield better allocation welfare. The code and artifacts are available at https://github.com/KurbanIntelligenceLab/waiting-buys-two-moments.

0.36

World of SWECraft: Agent Collaboration at Scale via Simulated Software Worlds

cs.AI cs.CL ·2026-10-16 ·Shannon Zejiang Shen, Yapei Chang, Qiyao Liang, Benjamin Newman, Jenny Liang +11 more
Abstract · 33 pages, 9 figures, 7 tables

As teams of AI agents take on larger projects and long-standing open problems, agent collaboration offers another dimension for scaling inference-time compute. To understand what makes such collaboration effective and scalable, we develop Software World, a framework for simulating multi-agent coworking in realistic software ecosystems. It comes with a pipeline to create environments of real, interdependent Python packages of different topics, sizes, and dependency structures, which we call software worlds. In a simulation, agents maintain packages in the world with a high-level goal of improving software efficiency, and they are free to discover tasks to work on and collaborate with others. We evaluate agents' work in a world collectively by measuring speedup in hidden dependent packages. The framework allows us to study different elements of collaboration by varying communication, ownership, and resource allocation. We run simulations in three different worlds, totaling more than 10,000 hours of agent work across eight models. Overall, performance improvement scales with inference-time compute, with the best model improving downstream efficiency by up to 31%. We show that most agents can find meaningful tasks via collaboration, but some frontier models can prematurely stop looking for improvements and make little progress. We release the simulation framework and traces to support future studies of agent collaboration at https://szj.io/sw.

0.35

Learning to Be Guided: Shaping Latent Geometry for Classifier-Free Guidance

cs.LG cs.AI ·2026-10-16 ·Haoming Liu, Junhao Zhu, Shaoyi Zheng, Yuanhe Guo, Jinnuo Liu +2 more
code/data
Abstract · 29 pages, 5 figures; project page: https://maps-research.github.io/cfg-geometry/

Classifier-free guidance (CFG) is a standard component of conditional diffusion models, and guided generation performance has become the \textit{de facto} measure of state-of-the-art models. Recent work on latent diffusion models seeks to learn latent spaces that are easier to model with diffusion. However, under CFG, generation additionally depends on how conditions are arranged in the latent space. We study this arrangement under discrete class conditions, termed \emph{class geometry}, and investigate what class geometry allows a latent space to work well with CFG. Under a Gaussian class model, we show that shifting variance from intra- to inter-class at fixed total variance makes classes more distinguishable under noise, and derive a noisy posterior objective whose per-sample descent direction is proportional to the oracle CFG correction. Based on the analysis, we propose Prototype Regularization (PR), a lightweight surrogate for the intractable noisy posterior objective, which applies a cosine prototype classifier to spatially pooled latents. With a controlled training budget (20 epochs for VAE, 80 epochs for DiT), adding PR to an LDM baseline reduces guided gFID on ImageNet from 3.31 to 2.28, surpassing a DINOv3-aligned VA-VAE (2.44) with negligible overhead. We further demonstrate that PR generalizes to pseudo-labels derived from clustering, continuous text-embedding conditions, and video latents. Together, our findings highlight class geometry as a critical factor in guided generation and take a first step toward CFG-compatible representation learning. Our project page is available at: https://maps-research.github.io/cfg-geometry/.

0.35

RL-Guided Temporal Localization for Dual-Channel Retrieval in Long-Horizon Agent Memory

cs.AI ·2026-10-16 ·Dabin Sheng, Zhe Wu, Jinming Zhao, Junliang Xing, Yuanchun Shi
top venue
Abstract · Accepted at NeurIPS 2026

Time-aware retrieval is crucial for long-horizon LLM agents to effectively leverage ever-growing long-term memory. Existing methods focus on lexical or semantic similarity, leading to target memories being crowded out by temporally mismatched yet semantically similar candidates. Even time-aware approaches inject temporal cues only during memory organization or answer generation, overlooking temporal constraints at the retrieval stage, where candidate selection occurs - the root cause of the displacement. Inspired by the temporal contiguity effect in human episodic memory, we propose TIDER (Temporal Interval-driven Dual-channel Evidence Retrieval), a time-aware agentic retrieval framework. Central to TIDER is a reinforcement learning (RL)-trained temporal interval locator that infers the temporal windows in which the evidence lies. RL optimizes window localization as a discrete, non-differentiable retrieval decision, using composite rewards to convert sparse answer supervision into fine-grained retrieval feedback. These temporal windows first define the temporal retrieval channel; the retrieved candidates are then merged with results from the semantic channel and reranked by proximity to the windows. This temporal grounding enables TIDER to retrieve temporally aligned evidence that would otherwise be drowned out by the volume of semantically similar yet temporally mismatched memories. With a lightweight 1.7B locator, TIDER achieves state-of-the-art accuracy on both long-horizon memory benchmarks - 61.52% on LifeBench (+8.82%) and 78.96% on LoCoMo (+3.66%) - which indicates that temporal cues are crucial for pinpointing target evidence amid a vast pool of semantically similar historical events in long-term daily-life memory.

0.35

Interpretability of AI weather models via intermediate decoding

cs.LG physics.ao-ph ·2026-10-16 ·Matthias Beylich, Kirsten I. Tempest, Tobias Selz, George C. Craig
Abstract · 16 pages, 10 figures, 1 table. Supplementary material in the ancillary files

AI weather prediction models are often regarded as black boxes. Despite their impressive skill, there is little understanding of how the input data is processed during inference to produce the forecast. This limitation arises from the use of high-dimensional latent representations in which individual dimensions are no longer directly interpretable in physical terms. To gain insight into how forecasts are constructed, we propose the method of intermediate decoding for AI weather models and apply it to GraphCast, Aurora and ArchesWeather. In this approach, latent states from intermediate depths are decoded back into meteorological fields. To do this, we use the decoder that has been trained to output the forecast at the end of processing. Since latent states at different depths do not share a consistent basis, we train affine transformations that align intermediate latent states with the decoder's expected input space. Applying this framework to the three models yields the following results. First, the skill of forecasts at intermediate stages of processing improves gradually through the processor layers. Second, bypassing all processor blocks shows that the encoder-decoder pair is not an identity mapping returning the input, but instead plays a significant role in the forecast construction. Third, by examining increments of individual processor blocks across model depth, we show that all three models progressively refine the forecast from coarse to finer spatial scales. Finally, we assess the physical consistency of the intermediate decoding forecasts and find that large scales remain physically balanced throughout processing, while small scales become more balanced with each additional processor block.

0.35

Distinguishability and Recovery for a Doubly Sparse Spiked Wigner Model

cs.LG stat.ML ·2026-10-16 ·Ioana Dumitriu, JD Flynn, Zhichao Wang
Abstract · 50 pages, 9 figures, 8 tables

We study distinguishability and recovery of a sparse rank-one signal in a spiked Wigner model with independently sparse noise through a variety of algorithms and parameter regimes. The signal and noise have separate sparsity parameters, while the signal-to-noise ratio (SNR) remains of constant order as the matrix size grows. For sub-Gaussian noise and spikes whose nonzero coordinates are bounded away from zero, a leave-one-out quadratic thresholding procedure (a generalization of Ding et al., 2024) identifies the support. Sign synchronization yields exact recovery for a Rademacher spike, and spectral refinement yields strong recovery. The quadratic thresholding search parameter gives an explicit tradeoff between signal sparsity, noise sparsity, and computational cost. We also study extensions to sub-Gaussian spike priors through diagonal and quadratic thresholding. For Rademacher spikes with continuous nonzero noise, we develop an identical search algorithm that exploits unblurred signal entries, the existence of which permits distinguishability; connectivity of the graph they reveal provides a sufficient condition for exact recovery. These guarantees do not require moment assumptions on the continuous noise distribution. Together with our previous PCA results (Dumitriu et al., 2026), our analysis covers almost the entire space of parameters and regimes for distinguishing and recovering in our doubly sparse spiked Wigner model.

0.35

GatewayBench: Unified Capability and Security Evaluation Platform for MCP Gateways

cs.AI ·2026-10-16 ·Yitian Zheng, Zhaorun Chen, Haoyu Li, Bo Li
benchmark
Abstract

AI agents have increasingly interacted with external environments by invoking tools, with Model Context Protocol (MCP) emerging as a standard interface for connecting agents to heterogeneous tools and services. As organizations deploy agents at scale, MCP gateways have become a critical middleware layer for aggregating, governing, and shaping the tool surface exposed to agents. Despite the growing importance of MCP gateways in agentic systems, no prior work has systematically studied how gateway design affects agent capability, efficiency, and security. To address this gap, we introduce GatewayBench, the first unified evaluation framework and benchmark for MCP gateways. We characterize gateway architectures into two major paradigms: flattened gateways and hierarchical gateways. GatewayBench comprises 524 tasks across six domains, together with a pool of 11,118 real-world tools from 251 authentic MCP servers. To emulate large-scale deployments, we systematically inject task-irrelevant tools sampled from this pool into each agent environment, enabling controlled evaluation as the accessible tool surface scales. Based on GatewayBench, we conduct extensive experiments across seven widely deployed MCP gateways and diverse agent configurations, covering both benign and adversarial tasks. Our results reveal substantial differences across gateway designs in task completion, tool-use efficiency, and resilience to attacks, and further demonstrate that these effects interact strongly with the underlying model and agent harness. These findings highlight the MCP gateway as a consequential yet underexplored component of the agentic AI stack. By enabling controlled evaluation of gateway architectures under realistic tool ecosystems, GatewayBench provides both practical guidance for MCP gateway design and a rigorous testbed for developing more capable, efficient, and secure agentic systems.

0.34

A Conditional No-Hair Theorem for Transformers: How Language Models Forget Their Prompt

cs.AI ·2026-10-16 ·Shuyang Xiang
theory
Abstract

We ask how much an autoregressive transformer retains of its prompt after generating $t$ tokens. Consider two different prompts followed by the same continuation text, so that any difference in their hidden states comes from the prompt alone, and we find that the distance between their hidden states decays approximately as a power law $t^{-α}$, where $α=1$ is the rate at which a fixed prompt's share of uniform attention shrinks and smaller values mean longer memory. Trained models forget fast ($α\approx0.55$--$0.7$), untrained ones slowly ($α\approx0.2$--$0.5$). We prove that this offset is $O\big(P(\log t)^{L-1}/t\big)$ for a prompt of length $P$ and depth $L$. The bound holds for finite-depth pre-LN softmax transformers with bounded attention logits, under one additional condition: the pre-norm activations stay away from zero. The mechanism is geometric: the prompt's direct influence is diluted as the context grows, and \emph{relay} --- continuation tokens passing on what earlier tokens read from the prompt --- can only add logarithmic factors. This is what we call the \emph{conditional no-hair theorem}: given the text generated so far, the prompt's influence on the next token vanishes as $t$ grows, so any property of the prompt that survives, such as its language, must be carried by the text itself. An exactly solvable linear relay model shows that the bound cannot be improved and reproduces the rise of the local exponent towards $1$ seen in untrained networks. Cancelling the dilution with a logit bias stops the decay early in training and reverses it after, so dilution is necessary for the forgetting we measure; untrained networks and the solvable model show it is also sufficient.

0.34

On the tradeoff between robustness and geometric accuracy in adversarial training

cs.LG math.AP ·2026-10-16 ·Rachel Morris, Ryan Murray, Kerrek Stinson
theory
Abstract

For non-parametric binary classification in the presence of an adversary, we prove an optimal quantitative estimate for adversarial classifiers controlling geometric accuracy, as measured by the Hausdorff distance to the Bayes classifier, in terms or their robustness, as encoded by the adversarial budget $\varepsilon$. This rigorously shows that robust classifiers sacrifice at most $O(\sqrt{\varepsilon})$ or $O(\varepsilon)$ in accuracy, depending on the regularity of the underlying Bayes classifier. In contrast to prior work on the incompatibility of robustness and accuracy over low-complexity hypothesis classes, our work shows that high-complexity hypothesis classes allow for robust classifiers with precisely quantified accuracy. Our results apply for adversaries determined by any norm on Euclidean space and in both the infinite and finite data settings.

0.34

Beyond Expected Values: Risk-Sensitive Planning with Distributional Monte-Carlo Tree Search

cs.LG stat.ML ·2026-10-16 ·Quynh Trinh, Trung Nguyen, Nam Nguyen, Tuan Dam
top venuetheory
Abstract · Proceedings of the 40th Annual Conference on Neural Information Processing Systems

Monte-Carlo Tree Search (MCTS) typically stores only scalar mean estimates, which makes it poorly suited for tail-risk objectives such as conditional value-at-risk (CVaR). We introduce Categorical Thompson Sampling with Optimism (CATSO) and Particle Thompson Sampling with Optimism (PATSO), two distributional MCTS algorithms that maintain finite-support empirical backup laws at Q-edges. CATSO represents Q-edge laws with categorical atoms and Dirichlet Thompson sampling, while PATSO represents them with adaptive particles. Both algorithms select actions using the CVaR of a Thompson-sampled Q-edge law plus a polynomial optimism bonus, and propagate scalar continuation values through visit-weighted averages of child Q-edge CVaR scores. For the finite-depth nested CVaR planning target, we prove an $O(n^{-1/2})$ bound on the root value error after $n$ rollouts, which upper-bounds the risk-sensitive simple regret. CATSO adds a fixed-grid discretization bias, while capped PATSO adds a tunable Wasserstein compression bias. Experiments on stochastic planning benchmarks show that distributional Q-edges improve lower-tail return and reduce catastrophic outcomes when mean-optimal and risk-sensitive behavior differ.

0.34

ENFR: Evidence Networks with Flow-Matched References for Amortized Simulation-Based Model Inference

cs.LG stat.MLastro-ph.CO ·2026-10-16 ·Kai Lehman, Jameson Dong, Carolina Cuesta-Lazaro, Niall Jeffrey, Benjamin D. Wandelt +3 more
Abstract · 28 pages, 6 figures

The evidence, $\mathcal{Z} = p(x | \mathcal{M})$, of a data-generating model $\mathcal{M}$ is a powerful tool in Bayesian frameworks: it enables model inference, model misspecification tests, and goodness-of-fit assessment as an alternative to standard discrepancy statistics, e.g. $χ^2$. While versatile, the evidence is notoriously difficult to compute, even when the likelihood is available. In both likelihood-based and simulation-based inference, estimating $\mathcal{Z}$ provides an immediate solution to high-dimensional goodness-of-fit or model misspecification tests: using $\mathcal{Z}(x)$ as the test statistic induces a minimum-volume level-$α$ test. Such tests are essential to assess model fidelity, which is necessary to trust simulation-based parameter inference. Existing learned approaches largely target Bayes factors, which only measure evidence ratios. Instead, we provide accurate, fast, and amortized estimates of the absolute evidence. We present ENFR: Evidence Networks with Flow-matched References. Building on the Evidence Network framework of Jeffrey & Wandelt (2024), the proposed method first learns Bayes factors with respect to a flow-matched reference, from which the absolute evidence is recovered. We show improvement in estimation of $\mathcal{Z}$ over the flow-based baseline in all applications considered and the best performance of all density estimation methods considered in our benchmarks. We furthermore introduce and provide the proof for an architecture that jointly learns the individual evidences of a set of models $\{\mathcal{M}_i\}$. We demonstrate the performance and scaling behavior of these approaches on data from a variety of scientific applications.

0.34

On the possibilities and limitations of LLM-based multi-agent systems for multi-criteria decision-making in healthcare operations management

cs.AI math.OCcs.MA ·2026-10-16 ·Marco Boresta, Tommaso Giovannelli, Sepehr Namirad, Giulia Riccardi
Abstract

Multi-criteria models for healthcare operations need weights that express how stakeholders trade off patient waiting time, provider workload, and operating cost, but these weights are difficult to quantify. Teams of large language model (LLM) agents, each prompted to represent a stakeholder, have been used to supply them. We propose a framework for assessing the reliability of LLM-generated preferences as inputs to multi-objective optimization and quantifying how their uncertainty affects decisions. We study explicit preferences by asking agents to rate the importance of the criteria in an outpatient appointment scheduling problem, and implicit preferences by inferring trade-offs from the agents' pairwise choices between emergency department configurations. The computational experiments show that the explicit preferences respond to the clinic's operating conditions and differ across roles, but deliberation among the agents can erase these differences, whereas simple averaging preserves them. The pairwise choices used to reveal the agents' implicit preferences depend largely on which option is presented first, and we find limited evidence that a statistical model using contextual information alongside presentation order predicts the agents' choices on new comparison pairs better than a model using presentation order alone. We then formulate robust weighted-sum optimization models to account for preference uncertainty. In most cases, these models select different schedules or configurations from those obtained using average preference weights. These results indicate that LLM-generated preferences need an audit of this kind before they are used as optimization weights in healthcare decision support.

0.33

lepidoptera: A Domain-Agnostic Engine for Causal Simulation with Mixed Variable Types, Vine Copula Dependence, and Individual-Level Counterfactual Propagation

cs.AI stat.MEcs.LGstat.CO ·2026-10-16 ·Danilo Augusto Sarti
code/data
Abstract · R and Python package available at https://github.com/danilosarti/lepidoptera

lepidoptera is an open-source R and Python package for simulation from structural causal models with heterogeneous variable types. The package is built around a typed mechanism architecture, which separates the causal graph from the generating rule at each node and exposes each rule as an inspectable, serialisable object. A pre-built mechanism library covers all nine combinations of continuous, categorical, and ordinal parent and child types, removing the need for users to implement type-specific sampling from scratch. A vine copula mechanism estimates the full joint dependence structure from observed data using pair-copula constructions and the Rosenblatt conditional-simulation algorithm, supporting over twenty copula families. Individual-level counterfactual queries are supported through partial propagation, which freezes the random draws of unaffected nodes and replays only the downstream consequences of an intervention, thereby separating the individual treatment effect (ITE) estimand from the population-level average treatment effect (ATE) without additional model assumptions. An extension layer allows any pre-fitted predictive model to serve as a generative mechanism via a unified callable interface. Three worked illustrations spanning clinical trial planning, emergency-ward capacity analysis, and epidemiological mediation analysis demonstrate the package's scope.

0.33

Robust Function Calling for LLM Agents Under Interface Drift

cs.AI cs.CLcs.SE ·2026-10-16 ·Aniruddha Chattopadhyay, Kanishka Halder, Partha Sarathi Roy
published
Abstract · 16 pages, 1 figure, 6 tables. Peer-reviewed and accepted at NGEN-AI 2026 (International Conference on Next Generation AI Systems). This is t…

Tool-augmented language-model agents rely on interface text: tool descriptions, argument schemas, and the candidate tool list shown at inference time. These interfaces often change after deployment even when the underlying tool semantics remain fixed, creating a practical reliability problem for function-calling systems. We study this failure mode as interface drift and evaluate whether a lightweight, training-free repair layer can recover robustness without retraining the base model. Our method, SchemaShield-Lite, combines schema validation, candidate-set retry for missing tool names, and one-shot constrained repair. On Qwen3.5-9B, it raises DICE-300 accuracy from 0.717 to 0.813 and BFCL-200 accuracy from 0.580 to 0.805, recovering 19/22 and 33/45 drift-induced failures with no observed repair harms. Naive retry does not explain the gains. Ablations show that candidate-set drift, especially distractor tools, accounts for most observed damage, while description and schema edits alone are less harmful. The results position validation-and-repair as a practical robustness layer for deployed LLM-agent tool use.

0.33

How Finite-Rank Bottlenecks Shape the Low-Rank Adaptation Landscape

cs.LG ·2026-10-16 ·Long Nguyen-Chi, Quynh Nguyen, Thanh Nguyen-Cung, Binh T. Nguyen
top venuetheory
Abstract

Low-rank adaptation (LoRA) has become the standard method for parameter-efficient fine-tuning of large pretrained models, yet theoretical explanations for why low-rank updates suffice remain incomplete. Most existing analyses rely on the Neural Tangent Kernel (NTK) regime, which linearizes the network around the pretrained weights. In this work, we take a different approach that requires no linearization: our key observation is that, once the backbone is frozen, the loss sees each adapted weight matrix only through a finite-rank linear map induced by the frozen activations. Trace-norm regularization then forces the optimizer to pick a low-rank matrix among all matrices that produce the same observed output. This yields global minimizers whose ranks are controlled by the rank of this map, which we call the \emph{bottleneck rank}, rather than the sample size. For multi-head attention, we obtain tighter rank upper bounds by exploiting two additional invariances: row-wise softmax is insensitive to additive row constants in the score matrix, and value updates are observable only after attention weighting and frozen output projection. For the factorized LoRA objective with weight decay, we prove a complementary result on the nonconvex side: after an arbitrarily small generic positive semidefinite perturbation, every first-order stationary point is rank-deficient once the LoRA rank exceeds a threshold determined by the bottleneck rank. Finally, we show that global minimizers of the perturbed objective are near-optimal for the original trace-norm problem, and we provide uniform generalization bounds governed by the spectral structure of the frozen features.

0.33

Response-Informed Value Estimation for Cost-Aware Multi-Turn LLM Routing

cs.LG ·2026-10-16 ·Haotian Yan, Jiaqi Liu, Bingyi Jing, Tianjun Yao
benchmarkcode/data
Abstract

Multi-turn LLM routing selects a model at every turn before its response is available, yet task success is revealed only when the interaction ends, leaving each intermediate choice without a label of its own. Logged trajectories, however, retain the response each call produced, which exposes what each turn required and how well the chosen model served it, evidence that outcomes alone hide. We introduce RIVET (Response-Informed Value Estimation for Turn-Level Routing), which exploits this post-call evidence during training while keeping routing strictly pre-call at deployment. RIVET jointly trains two value estimators over a shared state-model representation: a deployable Routing Value Estimator that scores candidate models from pre-call information, and a training-only Response-Informed Value Estimator that additionally observes the realized response. Conditioning on the response turns response-induced return variation from residual noise into an observed covariate, and we show that this reduces the leading asymptotic estimation error of the shared representation. On the MINT benchmark, RIVET solves more held-out tasks than the strongest multi-turn baseline while using 74.7% less total call cost.

0.33

MS-Eval: Benchmarking Sub-Agent Delegation and Orchestration in MCP-Intensive Working Scenarios

cs.AI ·2026-10-16 ·Bowen Ding, Qixin Wang, Lexing Ruan, Weiyi Lu, Qiwei Ge +2 more
benchmarkcode/data
Abstract · 48 pages, 20 figures

LLM agents are increasingly integrated into real-world works, yet they must handle diverse domain-specific tools and complex multi-sub-task problems that sequential tool invocation resolves inefficiently and with severe context bloat. The Main-Agent Delegation and Sub-Agent Execution (MAD-SAE) paradigm addresses these issues by delegating sub-tasks to specialized sub-agents for parallel execution in isolated contexts, but its delegation and orchestration quality remains unquantified. We introduce MS-Eval, a unified evaluation kit comprising an MCP-intensive benchmark, a controlled execution harness, tailored delegation and orchestration metrics, and the MSQ index for comprehensive assessment. Evaluating four MAD-SAE variants and eight frontier models reveals that (1) sub-agent specialization (Spec) is the best practical practice; (2) open-source models are largely competitive with few exceptions; and (3) MAD-SAE generally raises cost with contingent payoff, benefiting most open-source models in both performance and efficiency while leaving closed-source models with higher cost and degraded performance. Our case study further traces these gains to better single-agent solutions, effective context offloading, and genuine sub-task parallelism, whereas unnecessary or redundant delegation turns the same cost into pure overhead. Codes: https://github.com/LINs-lab/MS-Eval

0.32

Mesoscopic learning dynamics of extensive-width neural networks

cs.LG stat.MLcond-mat.dis-nncond-mat.stat-mech ·2026-10-16 ·Shu Zhou, Jean Barbier, Sebastian Goldt
Abstract

Understanding how stochastic gradient descent learns internal representations in wide neural networks trained on rich target functions is a central problem in machine learning. We study this question in two-layer neural networks learning target functions represented by networks of the same architecture, with arbitrary nonlinearity, and with network width and input dimension growing proportionally. In this regime, standard approaches for describing the learning dynamics require tracking a diverging number of overlaps, which measure correlations between learned and target features, leaving no obvious description in terms of a few order parameters. Here, we develop a mesoscopic description of learning by characterizing the distribution of overlaps. In the early stages of learning, we find that a Gaussian description of overlap statistics reduces the learning dynamics to a few collective variables that track learning of the target function's linear and quadratic components. We then estimate the timescale of specialization, when Gaussian closure breaks down and individual learned features begin to strongly correlate with specific target features. We find that specialization in extensive-rank networks is accelerated via a quadratic shielding mechanism, whereby learning the linear and quadratic components of the target function filters gradient fluctuations. We finally describe how the activation function determines the pattern of specialization and characterize the subsequent refinement of learned features. Together, these results show how collective learning gives way to individual feature discovery and clarify when learning in large neural networks admits a simple statistical description.

0.32

A Feature Selection Layer Approach for Multi-Class Classification on Neural Networks

cs.LG ·2026-10-16 ·João Pedro Silveira e Silva, Márcio Dorn, Bruno Iochins Grisci
publishedbenchmark
Abstract · 75 pages (including Appendix with Supplementary Material). Published in Information Sciences

In the age of data-driven decision-making, deep learning models have become powerful tools for uncovering patterns and making predictions that influence many aspects of society. However, their broader adoption in scientific and business applications is often limited by a lack of interpretability. Feature weighting methods, which assess the relevance of individual features, are among the most widely studied approaches for improving interpretability. While post hoc techniques have advanced significantly, especially in modeling nuanced relationships between features and labels in multiclass problems, embedded methods typically fall short in this regard. Embedded techniques integrate feature relevance learning directly into the model training process, but they are often unable to capture label-specific relationships. To address the limitation, we propose multiple extensions to the Feature Selection Layer, an embedded feature weighting mechanism for neural networks. Through extensive experimentation, we identify the most effective extension and benchmark it against established feature weighting methods. Our results show that the proposed method matches existing techniques in identifying globally relevant features and outperforms them in capturing label-specific relevance. By leveraging the internal dynamics of the trained neural network, it reveals how individual features contribute to the prediction of specific labels, requiring no post-processing beyond the training process.

0.32

SimOPD: Relational Knowledge Transfer for On-Policy Distillation

cs.AI cs.CL ·2026-10-16 ·Kai Tang, Xing Zheng, Haobo Wang, Guangcheng Zhu, Shenzhi Yang +4 more
Abstract

A key challenge in on-policy distillation (OPD) is extracting effective supervision from teacher knowledge. Supervision from token probabilities may underutilize internal knowledge, while direct hidden-state matching forces the student to shift to the teacher's feature space and complicates cross-architecture transfer. Inspired by contrastive learning's use of relationships between instances, we propose \textbf{SimOPD} to transfer teacher internal knowledge through response similarities. Specifically, SimOPD uses paired memory banks of student-generated responses to extract supervision from the teacher's internal representations as similarity distributions. The student then matches these distributions within its own representation space, retaining representational freedom and removing the need for cross-model projection. Extensive experiments show that SimOPD improves the overall average over the strongest baselines by up to 2.6 and 3.1 percentage points in same- and cross-architecture settings.

0.32

ToPA: Block-wise Toeplitz Adaptation for Expressive and Efficient Fine-Tuning

cs.AI cs.LG ·2026-10-16 ·Sicong Li, Qianqian Xu, Zhiyong Yang, Zitai Wang, Longtao Huang +2 more
Abstract

Parameter-Efficient Fine-Tuning (PEFT) has emerged as a prevalent strategy for adapting large pre-trained models. Among PEFT techniques, reparameterization-based methods, particularly those that employ low-rank or sparsity structures, have attracted significant attention. However, these approaches often underperform full fine-tuning. Specifically, low-rank methods constrain the update rank, which is misaligned with the inherently high-rank nature of full fine-tuning. In contrast, sparsity-based techniques limit the flexibility of the parameter space and compromise connectivity across weights. To overcome these limitations, we explore the use of Toeplitz matrices, whose entries are constant along each diagonal, therefore providing a compact parameterization without imposing explicit rank or sparsity constraints. A straightforward approach is to update weights via a product of Toeplitz matrices. However, similar to RNNs, long multiplicative chains often lead to gradient instability. To address this issue, we propose \textbf{ToPA}, a more refined version of this idea that replaces standard Toeplitz matrices with block-wise Toeplitz, greatly increasing each factor's capacity and shortening the chain, thereby stabilizing training and improving expressiveness. We theoretically show that products of block-wise Toeplitz matrices can approximate arbitrary matrices under shorter chain length, justifying the structural design of ToPA. Furthermore, we demonstrate that ToPA offers greater expressivity than existing low-rank and sparse parameterizations. Empirical evaluations across 21 NLP and CV datasets, spanning 5 model architectures, consistently validate the effectiveness of ToPA.

0.32

Annotrace: A Workbench for Step-Level Annotation of Agent Traces

cs.AI cs.CLcs.HCcs.MA ·2026-10-16 ·Chenyi Huang, Arun Josephraj Arokiaraj, Bin Wu, Sahan Bulathwela
Abstract · 4 pages, 4 figures, 1 table. Under review at the AAAI 2027 Demonstration Program

Every study of why LLM agents fail begins by labelling traces by hand, and every such study builds its own throwaway tool. Observability platforms record agent traces but offer no structured labelling; annotation tools label text but have no notion of a step, a delegation or a tool call, leaving it unclear what the smallest unit of execution to label should be. Annotrace reads six trace formats, including OpenTelemetry GenAI, LangSmith and Langfuse, and turns each run into an ordered, nested list of steps. The annotator writes their own labelling codebook; a frontier model proposes a starting set from a sample of their own traces, answers first, and is asked exactly the questions the human is asked, making annotation a joint human-AI process. Annotrace measures agreement between any two annotators, human or AI. On a multi-agent question-answering corpus it reproduces a previous manual effort's 235 annotation units exactly, and two models labelling the same traces agree 91% of the time at $κ= 0.82$.

0.32

LLM Evolution in a Simple Simulated Market

cs.MA cs.AIcs.CLq-bio.PE ·2026-10-16 ·Charlie Pilgrim, Christopher Krupenye, Dora Biro, Richard P. Mann
hot direction
Abstract

Arguments that AI systems will evolve in harmful ways are largely made in the abstract. We provide a concrete empirical example. We simulate simple markets consisting of populations of twenty Large Language Model agents tasked with solving arithmetic problems. Agents differ only in their system prompt; those that go bankrupt are removed and replaced by mutated copies of the prompts of successful agents. Two market conditions select for opposite phenotypes: a market that rewards accuracy produces agents with high accuracy (94%) and high token use; a market that charges per token produces agents with low accuracy (10%) and low token use. In the second market, only tokens are incentivised yet we see a collapse in accuracy. Token use and accuracy are correlated traits such that selection on one also moves the other, and behaviour adapts along an unpriced dimension. A third neutral condition with no selection allows us to separate these effects from bias in the mutation operator. Adaptation occurs through a collective optimisation process that operates above the level of individual agents, who are never told the payment rule, do not receive feedback, and have no memory.

0.31

ADT: All-Atom Diffusion Tokenizer for 3D Molecular Structures

cs.LG ·2026-10-16 ·Xuantuo Huang, Encheng Su, Aoran Wang, Pan Tan, Shengyong Xu +4 more
Abstract

Discrete structural tokenization provides a promising route to autoregressive 3D molecular generation. Yet, existing tokenizers face a trade-off between compact representation and all-atom reconstruction: atom-level tokens preserve atomic details but produce long sequences, while compact residue-level protein tokens typically omit side-chain geometry. To overcome this limitation, we propose All-atom Diffusion Tokenizer (ADT), which decouples token granularity from reconstruction resolution. ADT uses one structural token per atom for small molecules and one token per protein residue that jointly represents backbone and side-chain geometry. To mitigate geometric distortion introduced by quantization, we leverage the generative capacity of diffusion models to recover fine-grained atomic geometry beyond deterministic decoding from discrete tokens. Extensive experiments across small molecules and proteins demonstrate that ADT achieves strong reconstruction from compact discrete representations while enabling high-quality autoregressive generation of both small molecules and all-atom protein structures. These results establish compact structural tokens as an effective discrete interface for all-atom molecular reconstruction and generation.

0.31

Pareto-Optimal RTL Code Generation via Multi-Objective Reinforcement Learning with Large Language Models

cs.AI ·2026-10-16 ·Koki Takeshita, Kenji Uebayashi, Hikaru Tomonari, Masaya Tsunokake, Koichiro Yawata +4 more
publishedbenchmark
Abstract · Accepted at COLM 2026

Recent studies have proposed reinforcement learning with verifiable rewards (RLVR) methods that enable LLMs to generate high-quality RTL circuits from natural-language specifications. Existing approaches optimize circuit PPA (Power, Performance, Area) by aggregating multiple objectives into a single scalar reward, making it difficult to simultaneously generate diverse Pareto-optimal solutions across different trade-offs such as low power, low latency, and small area. However, in practical design, the relative importance of each objective varies depending on product requirements and design constraints; thus, obtaining a diverse and high-quality candidate set is beneficial in that designers can choose a preferred trade-off a posteriori. In this work, we are the first to propose a learning method for LLM-based circuit generation that optimizes, not the quality of a single circuit, but the quality and diversity of the entire circuit set obtained through multiple generations in the PPA objective space. The proposed method consists of (1) a generation-based training pipeline that reformulates coverage improvement of the entire Pareto solution set into a sequential circuit generation problem, and (2) a reward function based on Hypervolume Contribution (HVC) that measures how much a generated circuit contributes to the Pareto front. Evaluation on the RTLLM v2 benchmark shows that the proposed method achieves +13.3 percentage points in relative Hypervolume over a baseline without PPA optimization and +14.4 percentage points over a single-objective reinforcement learning method, while also demonstrating superiority in circuit diversity.

0.31

Client-Aware Security Adaptive Early Exit Vision Transformers for Heterogeneous Federated Learning

cs.LG cs.AIcs.AR ·2026-10-16 ·Sahar Refaei Ghalibafan, Samira Nazari, Christian Herglotz, Maksim Jenihhin, Mahdi Taheri
Abstract

Federated learning (FL) trains a shared model across data-owning clients without centralizing raw data, but client data difficulty and compute budgets can differ. We study local early-exit Vision Transformer (ViT) inference, where an intermediate classifier may return a prediction before the final Transformer stage. The proposed method replaces one global acceptance threshold with a client-specific base threshold from label entropy and shallow-exit confidence, then adjusts each input using predictive entropy, confidence, top-two class margin, and exit depth. Aggregation uses only clients that updated a parameter. Security-aware routing reaches 88.70% CIFAR-10 accuracy with 32.0% depth saving and 73.7% ImageNet-1K top-1 accuracy with 30\% saving. On matched six-client non-IID CIFAR-100, full-depth accuracy is 72.28% versus 68.20--71.80% for four heterogeneous-FL baselines; robustness-aware routing reaches 70.01% with 32.55% saving.

0.31

Persistent Memory and Iterative Improvement in an LLM Agent for Chinese-Character Drawing

cs.AI ·2026-10-16 ·Peilin Wu, Yi-Zhuang You
hot direction
Abstract

Does persistent memory accumulated from previous attempts help an LLM agent solve new problems, and does memory design affect that benefit? We study programmatic Chinese-character drawing across five agent conditions attempting the same 668 items. Accumulated memory produces modest aggregate gains despite substantial growth in the memory stores: free-form notes improve first-attempt success by 4.6 percentage points, while a design combining free-form memory with reusable code improves it by 2.8 points. The two designs do not differ significantly overall, but the code bank shows a suggestive advantage on later, harder items. A separate post-hoc comparison finds a much larger improvement when the code-bank agent receives privileged, target-specific structural supervision. The agent can therefore exploit highly informative structural guidance, although the privileged per-item information does not establish what could have been learned from experience alone. These results suggest that memory design may affect where accumulated experience becomes useful, even when its average benefit is modest.

0.31

"Deep lifts": Neural networks can generalize even with the dumbest optimizer

cs.LG ·2026-10-16 ·Sunil Madhow, Yu-Xiang Wang
Abstract

Given a labeled dataset $\{(x_i, y_i)\}_{i = 1}^n$, consider drawing a parameter vector $\hatθ$ uniformly at random from those that interpolate the data within a width-$W$ depth-$D$ ReLU architecture with continuous weights bounded within $[-M, M]$. We study the generalization properties of $\hatθ$ in the case where the dataset can, in fact, be explained by a smaller $w$-by-$d$ ReLU network. Despite the obliviousness of the selection rule for $\hatθ$, it adaptively generalizes at the rate $\tilde O\left(\frac{wdWD}{n}\right)$, rather than being governed by the VC dimension of the full $W$-by-$D$ hypothesis class. The reason is due to a particular inductive bias in ReLU networks: the $w$-by-$d$ parameter *witnesses* an exponential volume of $W$-by-$D$ interpolating parameters, which dwarfs the volume of poorly generalizing parameters (conditioned on interpolation). We explicitly construct this set of near-equivalent $W$-by-$D$ parameters by a proof technique called *deep lifting*, and then tie the achievable volume to stability properties of the $w$-by-$d$ witness. Examples show that these techniques allow us to convert approximation-theoretic results into generalization guarantees for nonparametric classes. We further show that generalization of uniform-random interpolators can survive label corruptions.

0.31

Ori: Drift-Aware Streaming $k$-PCA for Very Low-Rank LLM Training

cs.LG ·2026-10-16 ·Xinyu Luo, Jincheng Zhou, Mengbo Wang, Zilin Shen, Nadia Atallah Lanman +3 more
code/data
Abstract

A prominent family of low-rank LLM optimizers, pioneered by GaLore, performs gradient-subspace projection: at each step, they project the gradient into a low-dimensional subspace and map the update back to the full model. Most existing work either recomputes the gradient subspace at certain intervals, or performs post-hoc adaptation to catch up to the moving subspace. In this work, we explicitly identify in practice the slow, stable manner in which LLM gradient subspaces drift across time. To this end, we propose Ori, a novel low-rank optimizer for training LLMs that captures the shifting principal gradient directions over time to achieve state-of-the-art accuracy under stringent low-rank constraints. Ori employs a streaming $k$-PCA algorithm inspired by Oja++, whose low per-step cost enables high-frequency, on-the-fly subspace tracking. In practice, Ori reduces the performance gap of the state-of-the-art low-rank optimizers to full-rank optimizers by up to 69.3\% on a 1B LLaMA pretraining task, and achieves the best performance on 10 out of 16 task scenarios in Roberta-Base fine-tuning. Together, these contributions allow Ori to further push the performance and efficiency frontiers of gradient-projecting optimizers, narrowing the gap to full-rank training for LLMs. Code can be found at https://anonymous.4open.science/r/Ori-Submission-2132.

0.31

From SEC Filings to Financial Signals: Evaluating LLMs on Stock Market Reactions and Analyst Forecasts

cs.AI ·2026-10-16 ·Shengkun Wang, Taoran Ji, Linhan Wang, Sattar Mansi, Chang-Tien Lu +1 more
hot direction
Abstract

Form 10-Q MD&A sections combine long-form management narrative with accounting and forward-looking information. Benchmarks for financial filings commonly evaluate question answering or numerical reasoning, leaving unclear whether outputs from large language models (LLMs) align with market and analyst outcomes. We evaluate prompted, zero-shot outputs from 13 general-purpose and finance-specialized LLMs on 45,964 Form 10-Q filing events from 2001 to 2024, linking MD&A text to cross-sectional fundamental ranks, pre-filing analyst forecasts, and realized abnormal returns. The evaluation covers return ranking under five controlled information views and next-quarter EPS forecasting under direct and consensus-anchored interfaces. Relative to disclosure alone, adding cross-sectional fundamental ranks raises mean effective-date rank correlation (RankIC) by 0.0089 for effective-event-date returns and 0.0098 for following-day returns, whereas analyst-summary context provides no comparable average gain. Across the four disclosure-containing views, finance-specialized models show an effective-event-date group advantage but no statistically reliable following-day advantage, while general-purpose Qwen models produce the strongest individual return-ranking results. Under the full-information view, direct LLM forecasts average the 43.4th percentile of the analyst distribution and do not consistently beat individual analysts. Consensus anchoring raises the mean inside-range share from 41.1% to 59.5% but increases reported-EPS error.

0.31

CME-SpectrumBench: Can LLMs Analyze Condensed Matter Spectral Data?

cs.AI cond-mat.othercond-mat.str-elcond-mat.supr-con ·2026-10-16 ·Jin Gene Wong, Anjney Midha, Joseph Mathew Tennyson, Wei-Lin Chiang, Ion Stoica +1 more
benchmarkcode/data
Abstract

Large Language Models (LLMs) are presently well assessed by expert-level mathematics and physics benchmarks which focus on theoretical, symbolic reasoning. However, we lack experimental benchmarks that evaluate data analysis in frontier research problems, a capability that future domain-specific foundation models must have. To this end, we present CME-SpectrumBench, an expert-designed benchmark consisting of 190 questions that measures how reliably frontier LLMs analyze research-level spectral scientific data in condensed matter experiments (CME). Using simulated CME spectral data, we prompt models to identify structures from 2D intensity arrays in the face of noise and limited instrumental resolution, a test of dense context reasoning. Models are scored by how accurately their numerical answers approach the ground truth. Under direct query, scores generally pale in comparison to handwritten code due to clear failure modes: models can find simple features but struggle to trace patterns or compute derived quantities, and performance typically degrades with increased data resolution and noise/convolution levels. LLM performance becomes comparable to handwritten code by allowing tool use (access to a Python interpreter) with/without ReAct prompting, in effect mapping questions onto agentic code generation problems. Our findings show that frontier models can serve as research assistants by interfacing with data through code, but still face difficulties processing information directly from large datasets - a key barrier future foundation models must overcome. The code and data are publicly available at https://huggingface.co/datasets/jingene/CMESpectrumBench.

0.30

Can Compact Open-Weight Large Language Models Recover Mobility Events from Incomplete Trajectories? A Controlled Evaluation Using Air Trips

cs.AI cs.ET ·2026-10-16 ·Hsuan-Chih Wang, Xianfeng Terry Yang
Abstract

Incomplete mobility trajectories can obscure the observations needed to identify important travel events and their endpoints. This study evaluates whether compact, locally deployed large language models (LLMs) can recover missing mobility information from incomplete trajectories, using high-confidence air trips as a controlled test case, and whether more explicit prompts can compensate for the loss of endpoint evidence. Using nationwide location-based services (LBS) data from the United States in January 2022 and airport reference data, thirty high-confidence air trips were converted into three incomplete-trajectory conditions by removing observations near the origin airport, the destination airport, or both. Four prompt variants and two LLMs generated 720 requests, each using a two-stage procedure to identify the flight transition segment and then select the origin and destination airports. Exact transition accuracy is 9.2%, while origin and destination accuracy are 31.4% and 10.7%, respectively. Only 1.9% of requests recover both the rule-identified transition and the reference airport pair. The most explicit prompt produced the highest observed exact transition and exact airport-pair recovery rates. The study introduces a controlled endpoint-removal framework for evaluating recovery of discrete mobility events from incomplete trajectories, allowing failures to be examined by inference stage and helping analysts decide whether recovered information should be retained, independently verified, or left unresolved.

0.30

Unit-level aggregation breaks coverage in causal foundation models

cs.LG stat.ML ·2026-10-16 ·Gabe Schulman, Lucius E. J. Bynum, Yoav Wald
code/data
Abstract · 23 pages, 6 figures, 8 tables. Code: https://github.com/gabeschulman-nyu/subset-cate-coverage

Causal foundation models (CFMs) show promising results for treatment-effect estimation, with unit-level estimates that are competitive on standard benchmarks. Many causal analyses, however, inform decisions through effects averaged over groups of units, such as the full sample, the treated units, or a subgroup, and for such decisions a CFM is useful only if its credible intervals are calibrated. We examine these intervals and find that their coverage degrades, both in simple synthetic settings and on standard benchmarks. On ACIC 2016, for example, temperature-calibrated 95% intervals from CausalPFN fall below nominal coverage beyond groups of 30 out of 4802 units, and intervals for the full-sample average contain the true value in only 20% of datasets. We trace this problem to how these models are trained and how their intervals are formed. They output a predictive distribution for one unit at a time and, at inference time, combine the unit-level variances as if the errors on different units were independent. We show formally that posterior uncertainty over the data-generating process makes these errors correlated. As a result, the combined variance is too small by a factor that grows with the size of the group. Empirically, this factor predicts where coverage fails and by how much. We then show that two neural process models, which make joint predictions for a group of units, account for this correlation and substantially improve coverage on synthetic tasks and, as heads on a frozen CausalPFN, on all six benchmarks.

0.30

WebCoder: Scaling General Frontend Development with Artifact-Grounded Agentic Reinforcement Learning

cs.AI ·2026-10-16 ·Zijun Dou, Zhen Yang, Xijun Liu, Dongzhuoran Zhou, Aohan Zeng +1 more
code/data
Abstract · 50 pages, 22 figures, 11 tables. Project page: https://douzj-llm.github.io/WebCoder/

Real-world frontend development requires agents to compose reproduction, generation, modification, and repair over an evolving application state. Learning a general policy for this process requires translating heterogeneous execution feedback into task-aligned rewards and effective credit assignment across long sequences of interdependent actions. We introduce WebCoder, a general frontend development agent trained with artifact-grounded agentic reinforcement learning. Our key idea is to assess diverse development activities through their effects on a shared executable artifact and its satisfaction of task requirements. We build Webdev-World, an executable environment that integrates code editing, application execution, browser inspection, and on-demand skills while preserving project state across interactions. An independent verifier evaluates the final application against task requirements to provide terminal rewards, while changes in intermediate artifact progress redistribute the resulting rollout advantage across development segments. Context compaction and message-tree sample construction support training on trajectories beyond a single context window while avoiding repeated loss computation on shared generated history. We further introduce Webdev-Bench to evaluate composed development requirements within a single project. On Webdev-Bench, WebCoder-27B raises requirement satisfaction from 64.8 for its Qwen3.8-27B base model to 72.8, exceeding Claude-Opus-4.8 (65.1) and GPT-5.4 (59.6); on Vision2Web, it improves the average visual and functional score from 41.1 to 62.4, on par with Claude-Opus-4.8. Ablation studies confirm the contributions of artifact-grounded credit assignment and long-horizon training.

0.30

Not Every User Says It All: Training Tool-Using Agents Beyond Ideal Simulated Users

cs.AI cs.CL ·2026-10-16 ·Han Bao, Zheyuan Zhang, Kaiwen Shi, Yanfang Ye
Abstract · 22 pages, 11 figures, under review

Tool-using agents are trained and evaluated almost exclusively with strong simulated users who state every detail clearly, yet real users are often weaker. We construct weak users by masking sentences of a simulated user's replies, holding the task and the user's knowledge fixed, and cross-evaluate agents trained with strong and weak users on tau2-bench and ToolWOZ. Weak users lower the success of every policy, and training with them does not help: no policy trained with weak users, through fixed masks, mixtures or a curriculum, serves weak users better than the strong-trained policy beyond run-to-run noise, and prompting the agent to ask more does not help either. We propose StaGPO (State-Grouped Policy Optimization): the agent writes what the task needs, what is missing and how it is being filled, and GRPO uses this state to compare turns of different rollouts that reach the same situation, assigning step-level credit. Within the standard recipe of a small SFT cold start followed by RL, StaGPO makes weak-user training useful: on ToolWOZ, the Qwen3.5-2B policy trained with weak users reaches 76.8% with weak users, the highest of any Qwen3.5-2B policy we trained, 4.8 points above GRPO and 2.8 above the strongest baseline; it is also 5.4 points above GRPO with strong users, and generalises to an unseen mask rate; trained with strong users instead, it serves weak users at least as well as GRPO trained with weak users. On tau2-bench it scores highest with weak users for both models, by a small margin for Qwen3.5-4B.

0.30

Censored Flow Matching for Synthetic Time-to-Event Generation

cs.LG stat.MLstat.COstat.ME ·2026-10-16 ·Jianbin Tan, Xingran Chen, Kevin He, Zhenke Wu, Bhramar Mukherjee +1 more
Abstract

Generating time-to-event data is increasingly important in survival analysis, but it remains statistically challenging because event times are often censored and thus prevent flexible generative modeling. To fill this gap, we propose a novel method called censored flow matching (CFM), aiming at providing a flexible flow-based generative mechanism for synthesizing time-to-event data as well as learning their corresponding probabilistic densities. Unlike existing methods, CFM does not rely on survival likelihoods to handle censoring, but instead exploits the monotonicity property of differential flows to address censoring via pullback sampling. Under this framework, we implement an alternating procedure between pullback sampling and CFM steps, leading to a flow-based expectation--maximization (EM) algorithm tailored for time-to-event inference. Under mild distributional conditions, we establish the convergence of CFM to the true event-time distribution as the EM iterations proceed. We then apply CFM to generate time-to-event data and estimate their probabilistic densities, where the observed censoring time can be incorporated to provide truncated sampling or truncated density estimation for survival inference. Through extensive simulation studies, we demonstrate the benefits of CFM for complex and large-scale data generation. Finally, we apply CFM to analyze data from the Organ Procurement and Transplantation Network (OPTN) and the Medical Information Mart for Intensive Care IV (MIMIC-IV), showcasing its benefits for synthetic and predictive survival analyses in terms of data fidelity, predictive performance, and empirical privacy characteristics.

0.30

Towards High Semantic Fidelity: Hyperdimensional Symbolic Messages in Multi-Agent Communication

cs.MA ·2026-10-16 ·Shoucheng Song, Chang Yao, Hao Wu, Youfang Lin, Sheng Han +1 more
published
Abstract · NeurIPS 2026 Accept

Semantic fidelity is essential for multi-agent communication under non-ideal channels, comprising transmission fidelity and conversion fidelity. Yet existing numerical and neural message formats cannot simultaneously achieve both. To address this, we propose a novel message format-\textbf{H}yperdimensional \textbf{S}ymbolic \textbf{M}essages (HSM) and its corresponding communication framework \textbf{Herm}. For transmission fidelity, Herm suppresses semantic entanglement via an entity-attribute-value structured encoding pipeline. For conversion fidelity, Herm first leverages a receiver-centric ego-binding mechanism to align semantics. Then, a similarity-aware dual-branch fusion module is proposed to prevent critical semantic dilution. To realize explicit message-semantic conversion, we devise an iterative decoding scheme via algebraic inverse operations. Experiments exhibit that Herm outperforms baselines across various noisy scenarios while maintaining superior structural readability.

0.29

Self-Building Labs

cs.AI cs.MAcs.RO ·2026-10-16 ·Jiaru Bai, Yuchi Zhao, Han Hao, Elia Savino, Miroslav Bogdanovic +6 more
Abstract

Self-driving laboratories (SDLs) accelerate scientific discovery by automating experimental execution and optimization. AI agents can increasingly propose new experiments, but existing SDLs can only run the procedures they were built for. Adapting them to a new experiment still requires extensive manual engineering, which makes laboratory flexibility the bottleneck for scaling discovery. We introduce the self-building laboratory (SBL), which designs, configures, and operates its own laboratory setup given a high-level scientific goal. SBL starts from whatever already exists, from an empty room to a working SDL, and builds the missing capabilities: the lab layout, its digital twin, robot skills, and experimental workflows. We demonstrate the system working across different robots and setups, scaling from a single workstation to laboratories with several robots working together. We validate SBL's end-to-end capabilities through a real-world quantum-dot synthesis experiment. By shifting human researchers from manually configuring and programming device automation to specifying scientific objectives, SBL makes scientific automation more accessible, flexible, and scalable. Project website: https://self-building-labs.github.io/

0.29

MetaCurate: Evolving Curation Filters for High-Quality Agentic Coding Data

cs.LG ·2026-10-16 ·Donghoon Kim, Seunghyuk Oh, Kevin Galim, Minseo Kim, Minjae Lee +4 more
code/data
Abstract · Project page: https://furiosa-air.github.io/metacurate/

Supervised fine-tuning on agent trajectories is an effective way to improve coding agents, but selecting useful training examples requires more than identifying successful runs. A resolved patch may follow inefficient reasoning or unnecessary tool use, while a failed run may still contain useful intermediate steps. Training pipelines therefore rely on rules and qualitative assessments to filter trajectories, yet designing these filters typically requires substantial human expertise, and validating improvements through repeated training and evaluation is expensive. We introduce MetaCurate, a training-free framework that evolves executable trajectory curation filters using pairwise LLM judgments. Disagreements between the filter's rankings and the judge's preferences provide targeted feedback for jointly revising its rules, rubric, and aggregation policy. The resulting filter scores trajectories independently, so pairwise comparisons are required only during search. Starting from a blank filter, MetaCurate autonomously discovers a curation strategy that achieves competitive downstream performance without task-specific handwritten criteria. We validate downstream utility through controlled SFT experiments on our uncurated trajectory pool and the Nemotron v3 SWE dataset, with random selection and hand-authored filtering as baselines. In our experiments, MetaCurate achieves the highest mean SWE-Verified score among the compared methods, including hand-authored rule-based filtering and a training-loop baseline whose search costs 3.5x as much.

0.29

Adaptation of Computer Use Agents to New Software with User Manuals

cs.AI ·2026-10-16 ·Chulhoon Yoon, Kyungri Park, Jinyeong Yun, Daeyun Jeong, Sukmin Yun +1 more
code/data
Abstract · 21 pages, 3 figures, 14 tables. Code: https://github.com/ku-autoai/MAGE

Recent advances in multimodal foundation models have significantly improved the ability to operate real-world software through graphical user interfaces (GUIs) using visual observations and natural language instructions. However, maintaining strong performance on unseen applications remains challenging, as practical tasks often require long-horizon procedural knowledge and application-specific operation patterns. In addition, constructing executable tasks and collecting corresponding trajectories for training still heavily rely on human-annotated data, incurring substantial cost. In this paper, we propose MAGE (Manual-augmented computer use AGent Execution and learning), a framework that leverages user manuals as a scalable source of procedural knowledge to improve both inference and training. During execution, MAGE retrieves task-relevant manual content and uses a planner to generate explicit procedural guidance for CUAs. To further account for the discrepancies between manuals and actual GUI environments, such as version differences or user settings, MAGE also adapts the agent policy through environment interaction. During training, MAGE converts manuals into executable task-plan pairs and uses the plans to guide environment rollouts toward task-relevant states; the resulting trajectories are then used to optimize the policy without human-authored demonstrations. This unified design enables effective utilization of manuals for both task execution and training, reducing reliance on human-authored training data. Experimental results across six applications demonstrate improvements in task success rates over strong baselines, highlighting the effectiveness of manual-guided planning and policy adaptation in new software environments. We provide the code and reproduction scripts at https://github.com/ku-autoai/MAGE.

0.29

ScoutRoute: Efficient Mixture-of-Experts Routing Using Internal Language Model Signals

cs.AI cs.CL ·2026-10-16 ·Rajdeep Pandhare, Delzaan Bharucha
top venue
Abstract · 10 pages, 1 figure, 6 tables. Published in Advances in Artificial Intelligence and Machine Learning, 2026

Efficient expert routing is a fundamental challenge in Mixture-of-Experts (MoE) systems, particularly for heterogeneous collections of independently trained Large Language Mod els (LLMs). Existing routing approaches typically rely on jointly trained gating networks, prompt-based predictors, or computationally expensive multi-model inference, limiting their flexibility and efficiency. This paper presents ScoutRoute, a lightweight post-hoc routing framework that predicts expert competence using internal language model signals extracted from a single forward pass through an inexpensive scout model. Hidden-state representations are used to estimate cross-model competence through lightweight logistic probes, while logit-derived confidence measures are incorporated within a domain-aware routing strategy for efficient expert selec tion without modifying the underlying language models. ScoutRoute isevaluatedon fiveopen-weightlanguagemodels across fourbenchmark datasets covering mathematical reasoning, code generation, commonsense reasoning, and multidis ciplinary knowledge. Experimental results show that ScoutRoute consistently outperforms individual routing signals while requiring only a single scout-model inference. The proposed competence probes also achieve superior calibration compared with conventional confidence measures, demonstrating that internal language model representations provide transferable information for accurate and efficient expert routing across heterogeneous language models

0.29

From Approval to Abuse: Privilege Escalation Attacks on Harness-Protected Coding Agents

cs.AI ·2026-10-16 ·Xianglin Yang, Yechao Zhang, Jiahao Liu, Yulin Chen, Yuxin Cao +2 more
Abstract · Preprint

Coding agents use execution harnesses to constrain the consequences of prompt injection by enforcing permission boundaries. These systems deterministically intercept any out-of-sandbox action and prompt the user for approval. To reduce approval fatigue, users can approve allow-rules proposed by the agent, after which any future action matching the rule is allowed without a prompt. In this work, we show that this mechanism introduces a new risk: untrusted context can influence the permissions presented for approval, with consequences extending beyond the immediate task. We investigate privilege escalation targeting this permission-update process. Specifically, adversarial context planted in a repository can reliably steer an agent performing a routine task into requesting a plausibly justified, yet dangerously broad allow-rule. If the user approves the request, the saved rule can effectively blind the harness, enabling adversaries to execute malicious actions that would otherwise require approval, without triggering human oversight. Across production platforms, including Claude Code and Codex CLI, we demonstrate that agents successfully elicit the target allow-rules in 58% of experimental instances. Once acquired, these rules exhibit a massive semantic blast radius, covering 56.9% of category-specific capability test cases. To further quantify the downstream risk, we show that conditional on rule approval, our exploit achieves verified, silent goal completion in 47 of the 58 eligible instances (81.0%), undermining the safety promise of human-in-the-loop oversight.

0.29

LongForge: Measuring and Improving Long-Context Instruction Data via Evidence-Assertion Reasoning Graphs

cs.AI ·2026-10-16 ·Ziyang Chen, Junlong Jia, Xing Wu, Tinghao Yu, Songlin Hu
Abstract · Work in progress

Existing quality control for long-context instruction-tuning data is a one-shot judgment inside the generation pipeline that leaves only a scalar, revealing neither which document parts an answer uses nor how many reasoning steps it writes down; the data can thus be neither audited nor controlled. We formalize data quality as a property of the alignment structure among document, question, and answer, and build LongForge, which pairs an analyzer that measures long-context data with a synthesizer that generates it. The analyzer reverse-reconstructs each sample into an evidence-assertion reasoning graph and programmatically computes four dimensions on it: breadth, depth, correctness, and completeness. Every reading traces back to specific evidence spans and reasoning steps, making data auditable. Measuring seven mainstream datasets reveals two previously unquantified defects: breadth barely grows with document length, and depth is generally low, the two being nearly orthogonal. The synthesizer then embeds the analyzer into generation, gating on correctness and completeness and deferring breadth and depth to a training-time mixture over measured values, which makes the training distribution controllable. On two backbones, its two mixtures, LongForge-1 and LongForge-N, outperform the strongest baseline by 2.2-4.7 points on average over six long-context benchmarks, and in a controlled comparison that holds the gate and the candidate pool fixed, raising breadth and depth alone is worth a further 1.16 points, making the structural distribution of training data a lever on downstream quality rather than only a description of it.

0.28

Partial-Revelation Panel Forecasting: Predictive Information Gain, Allocation, and Coarsening-Consistent Operators

cs.LG ·2026-10-16 ·Bin Wen, Dai-Qiao Zhang, Tien-Ping Tan
Abstract · 27 pages, 2 figures. Code and aggregate result summaries included as ancillary files

We study partial-revelation panel forecasting: given $k$ of $n$ annotator votes on the previous dialogue turn, predict the full count distribution of the next turn's panel. This cross-turn predictive information differs from same-item annotation value. Under any nested revelation mechanism, the population decrement satisfies $V(k)-V(k+1)=I(Y;R_{k+1}\mid X_k)$; a fitted model also contributes adjacent excess-risk differences. Coarsened-completely-at-random (CCAR) masking supplies the exchangeable random-rater interpretation used in our operators. We define coarsening-consistent panel operators with a closed-form Dirichlet-multinomial realization, learned completion, and rater-count transfer. Against per-level lookup, cold-start count negative log-likelihood differences are -0.422/-0.166/-0.014 nats on IEMOCAP/M3ED/EmoWOZ. A held-out rater experiment measures same-item value: first and second additional votes reduce NLL by 0.583/0.221, 0.578/0.184, and 0.180/0.069 nats, respectively. Cross-turn and same-item gains are not interchangeable, and CCAR masks are controlled interventions rather than observed arrival processes. Consistency is auditable but does not guarantee predictive improvement; per-level marginalization can outperform a shared table. Target-free allocation beats uniform only where row-level signal is predictable: at budget one it recovers 83--93% of retrospective feature class headroom on IEMOCAP/M3ED but does not improve EmoWOZ. Gains are modest and corpus-dependent, with several small or null partial-revelation differences. Code and aggregate result artifacts accompany the preprint.

0.28

Bridging Compressed Communication and Local Steps via Spectral Method in Federated Learning

cs.LG ·2026-10-16 ·Dmitriy Bystrov, George Chirkov, Daniil Medyakov, Dmitry Bylinkin, Aleksandr Beznosikov
Abstract

Federated Learning (FL) enables collaborative training of complex models across large-scale networks of edge devices while keeping data localized. However, data heterogeneity and high communication overhead remain fundamental bottlenecks. Recently, the spectral optimizer Muon has demonstrated remarkable advantages over Adam-like methods in centralized training. While recent efforts have attempted to transition Muon to FL, naive integrations into traditional architectures fail short. Crucially, existing approaches do not support the the communication compression and local steps --- the two foundational techniques of efficient FL. To address this limitation, we introduce a novel spectral FL method that utilizes both local updates and compression. We establish rigorous convergence guarantees under mild assumptions, notably eliminating the restrictive bounded heterogeneity condition. Empirical evaluation on standard image classification benchmarks validates the efficiency and robustness of our approach.

0.28

Escaping Parameter Space: Tight Generalization Bounds via Representation Quality

cs.LG ·2026-10-16 ·Niclas Göring, Shuofeng Zhang, Branton DeMoss, Ard A. Louis
Abstract

We derive a tight generalization bound for deterministic neural We derive a tight generalization bound for deterministic neural networks based on their last hidden layer representations. By evaluating representation quality with a parameter-free $k$-nearest-neighbor ($k$-NN) classifier, our certificate avoids traditional parameter-space complexity measures, KL divergences to a posterior, weight norms, and compression. The bound is tight: trained from scratch, it reaches an 8.0 percentage point gap to test error on CIFAR-10 (ResNet-18) and yields the first non-vacuous certificates we are aware of on CIFAR-100; with frozen DINOv2 representations, it certifies at 1.4\% on CIFAR-10 and 23.1\% on 1,000-class ImageNet. It decomposes additively into three empirical, independently measurable quantities: local class geometry ($k$-NN decoder error), decoder-network agreement, and training stability. This allows the bound to act as a diagnostic tool: we demonstrate that actively destroying local class geometry eliminates generalization while leaving dataset memorization intact, identifying class geometry as a causally relevant mechanism for generalization. Finally, we show that learned representations are significantly less sensitive to weight noise than linear readouts, suggesting why parameter-space certificates often struggle to remain tight.

0.28

When Can You Trust an Agent's Own Verifier? Stress-Testing Counterfactual Rationale Verification Under Search Pressure and Oracle Access

cs.AI ·2026-10-16 ·Jonghyun Lee, Moonsu Kim, Sunghwan Park
published
Abstract · 17 pages, 4 figures, 8 tables. Accepted at the NeurIPS 2026 Workshop "Who Verifies the Agents? Toward Reliable Agent Development"

Automated agent design increasingly trusts a single score to decide whether a change is an improvement -- but a score optimized aggressively enough ceases to measure what it was meant to. When an agent also explains its design, that explanation is a second signal we can check without human labels: negate each stated claim, re-run, and observe whether performance drops. We ask whether such a verifier stays reliable once an agent optimizes against it. Our contribution is not a new verifier but a way to stress-test one: a pressure ratio $ρ$ measuring how much the verifier steers the agent's search, and four falsifiable probes covering four failure modes -- giving no signal, being too weak to matter, being gamed by stating fewer claims, and being queried before the agent commits. We apply the test to our own verifier, the Rationale Consistency Score (RCS), inside an LLM agent that designs machine-learning pipelines, with one generator model, six tasks, and three seeds. First, in our prior configuration the verifier barely affects the search objective ($ρ\ll 1$), partly because of an accuracy gate, so its robustness had not been tested; it still shapes the reported artifact through a final-selection tie-break that $ρ$ does not register. Second, a capability control shows the agent shortens its claim lists when instructed, yet search does not, even at $ρ$ over 100 times the prior level; since feedback omits claim volume and horizons are short, this null result is largely structural rather than evidence of robustness. Third, when the agent can probe the verifier before committing, it keeps claims that pass while held-out faithfulness falls -- a decline invisible to in-loop metrics, though partly confounded by how RCS scores small pipelines. We offer these as preliminary evidence and propose a reporting standard: audit a verifier with $ρ$ and these probes before trusting it.

0.28

BMuon: Budgeting Muon's Spectral Geometry for Structured Compressibility

cs.LG ·2026-10-16 ·Rida Assalouh, Tom Jacobs, Rebekka Burkholz, Yani Ioannou, Umut Şimşekli
Abstract · 54 pages, 6 figures

Muon has emerged as a strong optimizer for neural network training and admits an interpretation as steepest descent under a spectral-norm geometry. Yet we find that Muon-trained Transformers can degrade sharply when their MLP hidden channels are pruned after training. To reconcile Muon's optimization performance with structured compressibility, we introduce Budget Muon (BMuon), which modifies Muon's spectral geometry so that hidden channels across blocks compete for a shared allocation budget. The resulting oracle is exactly normalized steepest descent under a family of norms that interpolate between Muon's geometry and structured sparse geometries. Our principal sparsity control is an interpretable global channel-allocation budget, rather than a setting-dependent sparsity-penalty coefficient. For linear factorization models, we show that Muon admits minimum-cost exact representations with redundant hidden width, whereas under the BMuon oracle geometry and a sufficiently scarce budget, every minimum-cost representation is width-minimal. For practical training, we derive a minorization-maximization surrogate for the exact allocation problem. On ImageNet-1K with ViT-S/16, BMuon improves over Muon under gradual structured MLP pruning, and its learned allocations provide a more effective pruning criterion than weight magnitude. On a 1B-parameter LLaMA-style model trained on FineWeb, BMuon improves over Muon under post-training MLP channel pruning.

0.28

ASE: Attribution-Guided Semantic Evolution for Time Series Forecasting

cs.AI ·2026-10-16 ·Yuxuan Zheng, Zhaoyang Li, Purui Liu, Hengbo Xiao, Guannan He
Abstract · 11 pages, 3 figures

External text can supply predictive information absent from numerical histories, but its value depends on how it is represented, combined and deployed. We present Attribution-guided Semantic Evolution (ASE), a modular framework for constructing textual factors, evaluating their conditional contribution, and deploying validation-selected numerical--textual predictors. ASE supports task-specific implementations with lexical measurements, semantic primitives or evolved event/state factors, coupled to suitable numerical experts. Its feedback-based implementation uses conditional refitting diagnostics to guide factor revision and combination search. We evaluate ASE on the four text-associated TESS datasets and dynamic CSI 300 stock ranking. The four forecasting implementations obtain test MSEs of 0.001041, 0.022439, 0.333115 and 1.8150, each below the corresponding published TESS score. On CSI 300, the frozen text-fusion implementation improves mean RankIC by 0.000642 over its numerical expert and by 0.000701 over matched shuffled text, with positive differences in all five seeds. Controlled text perturbations support functional reliance on text correspondence, while incumbent retention prevents regressions in the measured development criterion. Together, the experiments characterize ASE at two levels: forecasting performance of task-specific implementations and the conditional contributions of individual modules.

0.28

Towards optimal selection schedules for online active learning

cs.LG stat.MLcond-mat.dis-nn ·2026-10-16 ·Hugo Cui, Francesca Mignacco, Francesco Mori
Abstract

We consider the problem of online linear classification, where a stream of unlabeled samples is available, and annotation is costly. We investigate the optimal query policy, that prescribes whether and when a sample should be labeled and used for a training step, or instead discarded, so as to minimize a given objective. We focus on the realizable case, under Gaussian design, in the high-dimensional regime where data dimension and number of unlabeled samples grow proportionally. We consider two objectives: 1) the generalization error under a labeling budget, and 2) a cumulative cost function where mistakes have unit cost and queries cost $Δ$. We first show that the myopic policies that greedily minimize at all times the increment in the objective correspond to either small or intermediate-margin policies, that prescribe the selection of samples whose margins fall within certain intervals that shrink with time. Within this class, we numerically ascertain the optimal schedules for the selection thresholds that minimize the final values of the objectives. Taking inspiration from the myopic policy for the cumulative cost, we design a very simple query policy schedule that we compare to other proposed heuristics, on synthetic data and real medical image datasets. Finally, we discuss myopic policies for two-layer neural networks, and exemplify how naive margin-based policies can cease to be optimal.

0.27

Peak-Memory Schedules You Can Trust: Exact Realization, Certified Optimality, and the Limits of Recomputation for Inference

cs.LG cs.DScs.PF ·2026-10-16 ·Ivan Fomin, Ivan Koryabkin, Boris Sheludko
theory
Abstract · 24 pages, 5 figures. Preprint

Real-time inference on small devices needs low activation memory without extra latency, which makes operator order the most valuable memory lever: it is free in time. We show that the usual way of deploying a better order, handing it back to the compiler and regenerating code, is unreliable. In PyTorch Inductor the regenerated program misses its predicted peak in 22 of 54 CPU configurations. It is worse than the stock program in 9 of them and in 19 of 50 on a GPU, and on the GPU it also changes numerical results. We instead profile the generated program, model its memory exactly from its own statements and their read/write effects, and permute those statements. We prove that this realizes the modeled peak whenever allocation does not depend on history. It does so to the byte, with bitwise-identical results, in all our CPU runs. On the GPU we characterize why the caching allocator breaks the condition. To know when to stop, we give forward and reverse "trap" lower bounds, each computable with $O(n^2)$ minimum cuts. They certify the found order optimal for 94% of held-out wrapper graphs and all 16 microcontroller-class models we study, in under a second. So most inference graphs are already scheduled optimally. Branching networks are the exception, where reordering is free and pays, e.g. 10.0% on DeepLabV3. In compiled training we save up to 7.6%. Beyond the certified optimum we study selective recomputation with per-operator caps. It is strongly NP-hard for every fixed cap even at treewidth two, encodes knapsack exactly on series-parallel graphs, and is solvable exactly on bounded treewidth. It is useless on public tiny models whose peak is a single layer, but on architectures that hold a cheap value for a late reuse it pays: on the public ReDimNet speaker networks, whose order is already optimal, it lowers the peak by 5.2-6.1% at under 0.4% extra MACs, with bitwise-identical outputs.

0.27

Predictable Scaling: Efficient Model Exploration via Scaling-Law Prediction

cs.LG ·2026-10-16 ·Aishwarya Wesanekar, Sheng-Tao Yang, Jiawei Zang, Flavio Sales Truzzi, Chen Chen +18 more
hot direction
Abstract

Evaluating candidate changes (hypotheses) to recommendation models requires full-scale training runs consuming thousands of GPU-hours, severely limiting the hypothesis space engineers can feasibly explore. We introduce Predictable Scaling (PS), a systematic framework that exploits empirically verified regularity conditions to predict full-scale model performance from small-scale proxies. PS constructs scaled-down variants via composable scaling functions along distinct resource axes, and projects metrics to the target scale via curve fitting with calibrated uncertainty estimates. We compare three prediction strategies of increasing expressiveness: even the simplest (raw proxy) achieves prediction error within the empirical noise floor of repeated-run variance; the most expressive (Gaussian Process) provides an indicative 15% relative MAE improvement and calibrated uncertainty. Evaluation on two ranking models of increasing complexity---a task-specific model and a foundation model---demonstrates that PS correctly ranks actionable hypotheses across diverse hypothesis types at 79-88% compute reductions.

0.27

An important Bottleneck in LLM Optimizers: Principle-Level Local Optima

cs.LG ·2026-10-16 ·Nian Ran, Tianhe Jia, Zhengyang Mu, Yue Wang, Sebastian Brulin +4 more
benchmark
Abstract · 10 pages under review

Feedback-driven LLM optimizers are increasingly used for evolutionary search, Bayesian optimization, heuristic design, and scientific discovery. Yet rapid initial gains often give way to prolonged plateaus despite substantial remaining evaluation budgets. We investigate \emph{principle-level search collapse}, where an optimizer's search becomes confined to a narrow set of design principles, even though it may continue to generate distinct candidates and achieve small improvements. We assemble a diagnostic benchmark spanning molecular optimization, network sentinel-node selection, and GPU-kernel optimization, and introduce a decision-tree protocol combining principle narrowing, prolonged performance plateaus, and gaps to attainable reference solutions. Across 161 assessable runs on the three problem domains, 78.3\% receive support for collapse, including 55.3\% with strong support. Candidate-level variation can nevertheless conceal restricted search at the principle level. We then test \emph{diversity injection}, a common response to search stagnation, such as higher t, multiple LLMs, and web search, and show that it provides limited and inconsistent benefits: among 65 molecular runs, still 73.8\% and 53.8\% receive support and strong support respectively. We finally test reference-principle interventions and find that they improve some runs, but their benefits vary across models and tasks: access to a better principle does not guarantee its effective use. Our benchmark and diagnostic protocol provide a basis for evaluating future optimizers on principle-level collapse. Our findings highlight the need for online collapse detection and better mechanisms for generating and maintaining alternative principles during optimization.

0.27

Deep LLM layers expose passage-level context in EEG during silent reading

cs.LG q-bio.NC ·2026-10-16 ·Ingo Marquardt, Anthilia Alchanat, Priyanka Jain
Abstract · 41 pages (including 23 pages of Supplementary Material), 12 figures (including 5 in Supplementary Material)

In a previous report we showed that a decoder can recover open-vocabulary word-level information from 19-channel EEG recorded during silent reading, by aligning short EEG windows with hidden-state embeddings of the presented word taken from a large language model (LLM). That analysis contrasted two of the model's 33 layers: the context-free input embedding and one mid-depth layer. Here we ask how EEG-decodable structure is distributed over the full depth of the language model. We trained one decoder per layer of Llama-3.1-8B, with and without a causal transformer over the sequence of EEG windows. We distinguish word-level decodability from a context-tracking component. Word-level decodability peaks in early layers and declines steadily towards the last layer. In contrast, the context-tracking component of the retrieval gain rises with depth, both with and without a causal transformer. Unexpectedly, a decoder that sees a single 0.5-s window at a time tracks the text passage too. Two new controls dissect this component. A target-geometry control, a synthetic predictor made of the true target plus noise calibrated to each decoder's word-level accuracy, shows that context tracking is not a by-product of word decoding and target geometry. A control that permutes predictions across runs from either the same or from different sessions shows that most of the observed context-tracking is not caused by an EEG recording session signature. Instead, context tracking mostly follows the text. A single EEG window therefore carries passage-level information that a contextual target can absorb. The results locate word identity in the early layers of the language model, discourse state in the deeper layers, and make the choice of target layer a design dial for EEG language decoders.

0.27

DiffusionRVGE: Reinforcement Learning via Regression-Based Value Gradient Estimation

cs.AI ·2026-10-16 ·Jiayuan Wang, Tongyao Pang
code/data
Abstract

Reinforcement learning improves diffusion models using rewards on generated images, but repeated denoising and gradient estimation make training expensive. We introduce \method, an efficient method that constructs useful update directions by regressing centered rewards on centered final denoised images. It requires neither reward derivatives nor adjoint backpropagation through the sampling trajectory. A unified first-variation formulation identifies the value-gradient target with or without a trajectory-KL penalty. A population analysis under Gaussian noising explains the regression geometry, while a subspace analysis motivates clean-image coordinates. The implementation adapts rollout bandwidth and query weights to reward continuity: local rollouts suit smooth rewards, whereas broader groups provide informative contrasts for discrete rewards. Trained jointly on five rewards, one SD3.5-Medium model matches or exceeds DiffusionNFT's joint model on seven of eight metrics, reaching GenEval $0.94$, OCR $0.92$, and PickScore $23.82$ at about one fifth of its estimated training compute. In single-reward training, \method reaches matched reward thresholds $2.1$--$6.7\times$ faster than reproduced DiffusionNFT on identical hardware. Ablations support the clean-image estimator, direct direction transfer, and reward-dependent rollout choices. Code is available at https://github.com/jy-wang-tsinghua/DIFFUSIONRVGE.

0.27

GEMS: Molecular Structure Elucidation as Learned Search over Isomer Graphs

cs.AI ·2026-10-16 ·Utku Umur Acikalin, Yingheng Wang, Goncalo J. Gouveia, Tyler J. Schwertfeger, Frank C. Schroeder +1 more
Abstract · 36 pages, 9 figures, 17 tables

Identifying small-molecule structures from molecular formulas and tandem mass spectra remains a bottleneck in untargeted metabolomics and natural-product discovery. Even when database retrieval or de novo prediction misses the exact structure, the resulting candidates can provide starting points for exploring nearby structures through a few bond edits. We introduce GEMS (Guided Editing of Molecular Structures), which refines candidates through learned 2-switch bond edits that preserve the molecular formula and each atom's total valence. GEMS uses MIST, a pretrained spectrum-to-fingerprint model, to predict fingerprint probabilities once per spectrum. A neural policy compares each candidate's fingerprint with these predictions, links present features to the atoms and bonds that generate them, and scores all valid edits in one forward pass. The policy is trained on shortest edit paths between structures, so it can be pretrained on PubChem isomers without spectra and then fine-tuned with spectrum-predicted fingerprints. At inference, GEMS searches from multiple initial candidates matching the supplied molecular formula and ranks candidates by fingerprint agreement and multi-start consensus. With the target structure excluded from PubChem initialization, three edit rounds achieve Hit@1/Hit@10 of 29.64%/50.81% on NPLIB1 and 30.22%/41.33% on MassSpecGym, exceeding published baselines under comparable settings. On NPLIB1, GEMS achieves higher identification accuracy in about 3.3 minutes end to end, compared with 26.2 hours for our FRIGID run. GEMS also improves identification from generated candidates, including those refined by FRIGID. Experiments with exact target fingerprints identify fingerprint prediction error as a bottleneck in target recovery and final ranking. Deeper search recovers additional targets without retraining, with larger gains from weaker initial candidates.

0.27

Causal Value Attribution of Marketing Labels in LLM-Generated Ads: A Deconfounder Approach

cs.AI cs.CY ·2026-10-16 ·Danhong You, Wei Jiang, Zhiqiang, Zheng
Abstract · 74 pages, under review in Journal of Marketing Research

As AI-generated content becomes commoditized, economic value increasingly shifts from final artifacts to the upstream expertise used to specify inputs that steer LLM generation. Yet organizations lack methods for valuing this expertise and identifying which individual skills or configurations causally improve performance. This study examines generative marketing, where structured expert labels serve as skill primitives conditioning LLM-generated advertisements. Estimating their causal effects is difficult because outcomes reflect stochastic LLM mediation, high-dimensional label interactions, and multimodal confounding. This study proposes D-VAE (Deconfounder Variational Autoencoder), a counterfactual-inference framework integrating a multi-treatment deconfounder with causal representation learning. D-VAE recovers causally disentangled expert-labeling logic and derives PAC-based bounds on omitted-variable bias from single-treatment confounders. Applied to 29,177 foundation-makeup ad creatives, D-VAE outperforms state-of-the-art benchmarks. A field experiment shows that recommended configurations increase engagement by 32.7% and perform statistically indistinguishably from human experts, while a simulation study confirms causal attribution under complex confounding and treatment interactions. Overall, the study shifts value assessment from AI-generated products to skills, providing a foundation for pricing and optimizing AI-augmented skill portfolios.

0.27

TALLOP: Joint Optimization of Local Training and Communication Compression for Time-Constrained Federated Learning

cs.LG cs.DC ·2026-10-16 ·Honoka Anada, Yafei Ou, Shinya Takamaeda-Yamazaki
published
Abstract · Accepted for publication in The 32nd IEEE International Conference on Parallel and Distributed Systems (ICPADS 2026)

Minimizing the wall-clock time required to reach a target accuracy is crucial in federated learning (FL), where many applications require frequent model updates as data distributions evolve over time. Nevertheless, in heterogeneous environments, slow clients with limited computational and communication capabilities can delay global synchronization, forcing faster clients to wait and increasing the overall training time. To mitigate this straggler effect, recent studies have introduced per-round time constraints and adapted client workloads within the available time. However, these approaches primarily focus on local computation, whereas communication can become a major bottleneck in communication-limited environments, motivating the joint optimization of training workload and communication compression. Although previous studies have explored such joint optimization, determining an effective training--communication balance under a fixed per-round time budget remains challenging, as the resulting model performance depends on both training progress and compression-induced degradation. We therefore propose TALLOP, which jointly optimizes the local training workload and communication compression under per-round time constraints. TALLOP predicts the training loss achievable under different computation budgets from the early-stage mini-batch loss trajectory (LENS), and estimates the degradation caused by communication compression using a compression-error model (CLEN). TALLOP combines these predictions to select the training-communication allocation that minimizes the post-compression loss under the given round time constraint. Experiments in heterogeneous FL environments show that TALLOP generally achieves competitive or improved time-to-accuracy compared with approaches that optimize only local training or neither, although the benefit over training-only adaptation depends on the setting.

0.27

Non-Parametric Optimization for Scalable Learning in Stochastic Decision Problems

cs.LG ·2026-10-16 ·Mohsen Amidzadeh, Lauri Viitasaari, Mario Di Francesco
Abstract

Stochastic optimization (SO) plays a central role in decision-making under uncertainty. Among SO problems, time-varying stochastic optimization (TV-SO) is particularly important due to its applications in adaptive control and machine learning. Non-parametric approaches have been proposed for time-varying deterministic optimization; however, they have not been developed for their stochastic counterparts. This work addresses that gap by developing a stochastic variational framework based on Malliavin calculus. This framework yields non-parametric optimality conditions for SO problems with stochastic decisions and supports the design of a scalable deep-learning algorithm that is insensitive to the parameterization dimension. This algorithm, called the Stochastic Path Follower (SPF), is applied to two important problems under distribution drift, namely least-squares recovery and logistic regression. Experimental results show that the proposed approach outperforms state-of-the-art learning-based and gradient-based methods in both performance and scalability.

0.27

Benchmarking and Enhancing Agentic Causal Discovery

cs.LG ·2026-10-16 ·Pingchuan Ma, Rui Ding, Shuai Wang
benchmark
Abstract

LLM agents can now run causal discovery algorithms on a dataset and report what comes out. Whether they do more than that, such as noticing where an algorithm went wrong and fixing it with further tests, is hard to tell from current benchmarks. Most of them give the variables meaningful names, and an LLM shown only the names of the Tübingen cause-effect pairs already gets 99% of the directions right. With the names removed, the data are usually clean enough that picking a better algorithm than PC closes most of the gap. We introduce Heuresis, a benchmark of anonymous simulated datasets whose mechanisms range from linear Gaussian to heavy-tailed. For each family of datasets it separates what a better existing algorithm can gain from what only a closer look at the data can gain, and it traces every edge an agent changes. Every family leaves a gap that none of the algorithms, the router and the two causal-discovery foundation models we ran can close, so a score above the best chosen algorithm has to come from using the data. We instantiate the benchmark with an agent harness and three LLMs, GPT-6.1 Sol, Claude Opus 5.5 and Qwen 3.8 Flash. GPT-6.1 Sol surpasses the best chosen algorithm on all six families, by 14% adjacency F1 on the most difficult one, and Claude Opus 5.5 does so by 7.4% there. A third of the true edges the two agents restore on the two hardest families appear in no algorithm's output. The code and data will be publicly released to facilitate future work.

0.26

Look Back, Correct Forward: Error Feedback Calibration for Time Series Forecasting

cs.AI ·2026-10-16 ·Haonan Yang, Yulin He, Zhuo Li
Abstract

Time series forecasting models are increasingly deployed in continuously evolving environments. After a forecast is issued, future observations gradually arrive and reveal how the model has failed. Existing methods mainly treat such observations as supervision for parameter updates. However, it remains unclear whether verified forecasting errors can themselves provide causal feedback for correcting future predictions. To address this, we introduce Error Feedback Calibration (EFC), a fully test-time framework for deterministic post-hoc correction of time series forecasting. EFC constructs a Causal Error Diagonal (CED) from revealed errors of overlapping historical forecasts, providing a fresh view of the frozen forecaster's recent failure behavior. Then Cumulative Error Transport (CET) maps this error state to future residuals using cumulative estimation, while temporal out-of-sample validation controls the correction strength. Temporal Residual Learning (TRL) further predicts only the residual structure left after CET from recent CED trajectories, rather than relearning the entire forecasting error. Extensive experiments across forecasting backbones and datasets show that EFC broadly improves forecasting performance while substantially controlling harmful over correction.

0.26

Beyond Reward Scoring: Online Multi-Agent Preference Learning for Multimodal Response Generation

cs.AI ·2026-10-16 ·Ziyun Zou, Jiwoo Hong, Yifan Lu, Shao Tang, Weiyi Wu +3 more
Abstract · 32 pages, 4 figures

Multimodal preference alignment often follows a score-based RLHF pipeline: a reward model is first trained to learn human preferences, then used to score on-policy responses and optimize the response model's generation. This turns preference supervision into a scalar signal for generation: a model may learn to produce answers with higher preference scores, but is rarely trained explicitly to learn the reasons underlying human preferences. We therefore ask whether making preference understanding an internal capability of the response model can improve its generation. To investigate this question, we propose \textbf{PrefMind}, an online dual-agent reinforcement learning framework with two trainable agents: a response Generator and a preference Critic. Given a pair of candidate responses, the agents analyze and discuss which response better aligns with human preference. Human labels supervise their judgments and corrections. For pairs on which both the Generator and revised Critic correctly identify the human-preferred response, the Generator produces an improved answer, and the Critic compares it against the original preferred response to provide a generation reward. Both agents are jointly updated from interactions sampled with their current policies. Through this process, \textbf{PrefMind} trains the Generator to understand human preferences via explicit judgment and correction, and to translate this understanding into better responses. Experiments across three model scales show improvements in both standalone preference judgment and downstream response generation.

0.26

When Magnitude Suffices: Reconstructing Time-Domain-Gated Microwave Signals Without Phase on a Thoracic Phantom

cs.LG physics.app-ph ·2026-10-16 ·Maciej Cichoń, Patrycja Gołaś, Maksymilian Bartosik, Angelika Walczak, Agata Bielecka-Dąbrowa +1 more
unusual
Abstract · 18 pages, 7 figures, 4 tables

Time-domain gating separates the arrivals that form a microwave S-parameter measurement. It requires the measurement of complex values, so this application almost always uses a vector network analyzer (VNA). We show that, on one calibrated measurement configuration, the gated magnitude the VNA would give can be reconstructed from a magnitude-only sweep of a spectrum analyzer with a tracking generator (SA). We select the gate center for separability of the measured states, correct each instrument once by a per-bin subtraction of a reference state, and train a fully connected network on sweeps from both instruments. The test object is a torso phantom with lung-shaped compartments whose filling takes eleven dielectric states, each held out in turn. On $S_{11}$ the reconstruction from SA input reaches a shape error of 0.08 dB$^2$ with a state ordering of 0.81. This matches the reconstruction from VNA input. On $S_{21}$ it reaches 0.82 dB$^2$ and keeps the ordered groups of the states. An explicit phase prediction performs much worse, because gating needs the phase to about 0.2 rad RMS. Across eleven measurement configurations the dependence of phase on magnitude is strong for reflection and weak for transmission. This explains the difference between $S_{11}$ and $S_{21}$. Where the magnitude carries enough of the phase, a scalar instrument gives the gated response without any phase measurement after installation. On the open UM-BMID breast data set, with whole phantoms held out, the shape error of the network is 30 percent lower than that of a per-bin linear map.

0.26

Path to Prior-Data Fitted Relational Models via Learned Latent Multi-table Representations

cs.LG cs.AI ·2026-10-16 ·Mohammad Sadeq Abolhasani, Viswanath Ganapathy
published
Abstract · Presented at the Fifth Learning on Graphs Conference (LoG 2026), Boston, MA, USA, November 20-22, 2026

Prior-data fitted networks (PFNs) show that pretraining on large populations of synthetic tasks can enable generalization to unseen datasets with little or no downstream fine-tuning. This paradigm is effective for transformer-based tabular and relational foundation models such as RDB-PFN, but remains largely unexplored for graph-based relational models built on message passing. We pretrain the graph-based relational foundation model Griffin on large collections of synthetic relational databases and ask whether curriculum learning, which benefits transformer-based relational learners, also improves message-passing representations. Rather than fine-tune on downstream tasks, we use the pretrained encoder as a general-purpose relational representation learner and pass its latent representations to a frozen in-context predictor (\ticl{}); no component is trained on the target task. On 16 real-world relational classification tasks, evaluated over five seeds against raw-feature and untrained-encoder baselines, a purely synthetic curriculum that begins with schema-guided data and continues with schema-agnostic data gives the strongest frozen in-context performance\rev{, outperforming real-data-only pretraining and every unordered single-stage mixture of the corpora; when appended after real-data pretraining, it yields the best encoders overall}. These results provide initial evidence that PFN-style synthetic pretraining transfers to graph-based relational foundation models.

0.26

Information-Geometric Blind Spots of Conformal Test Martingales

cs.LG stat.MLmath.ST ·2026-10-16 ·Johan Hallberg Szabadváry
unusual
Abstract · 9 pages, 5 figures

Conformal test martingales (CTMs) are e-processes for sequentially testing exchangeability via p-values produced by a conformity function $A$. However, any distribution shift that preserves the push-forward $P\circ A^{-1}$ leaves the p-value distribution invariant, rendering it invisible to every CTM based on $A$. We formalise this blind spot through the \emph{$A$-equivalence class} $[P]_A$ and show that, for any target distribution $T$, there is a unique closest element $Q^\ast\in[P]_A$ in KL divergence---the \emph{$A$-cryptic I-projection}---with explicit density $q^\ast(z) = t(z)\,p_S(A(z))/t_S(A(z))$ and a Pythagorean decomposition into detectable and cryptic components. When $A$ is the Neyman-Pearson likelihood ratio, the cryptic component vanishes, connecting our result to optimal e-value design. We demonstrate the theory on high-dimensional image data by empirically constructing this I-projection. By conditionally resampling EMNIST letters to identically match the conformity scores of MNIST digits, we demonstrate that a distribution can be shifted to a disjoint support while leaving the test martingale mathematically blinded.

0.26

Explainable Machine Learning Analysis of Temperature Effects on Mechanical Strength of Fiber-Reinforced High-Performance Concrete

cs.LG ·2026-10-16 ·Jagaran Chakma, Zhiguang Zhou, Badhan Chakma, Priyanta Chakma
unusual
Abstract · 29 pages

Predicting the compressive strength of fiber-reinforced high-performance concrete subjected to temperature exposure is challenging because to the intricate interdependencies of mixture composition, fiber properties, curing conditions, and temperature. A data-driven framework to predict compressive strength utilizing 1,204 experimental records containing a diverse array of material designs and exposure situations. This study evaluates eight machine-learning models and demonstrate that ensemble approaches routinely surpass individual learners. Both gradient boosting and extreme gradient boosting exhibit superior performance on unseen data, each attaining a R^2 of 0.903. Gradient boosting produces the minimal root mean square error (6.67 MPa), whereas extreme gradient boosting achieves the lowest mean absolute error (3.97 MPa) and mean absolute percentage error (10.60%). The models are further interpreted utilizing Shapley additive explanations and partial dependence analysis, which reveal temperature as the primary factor influencing compressive strength, succeeded by fine aggregate, superplasticizer, cement content, water-to-binder ratio, and water content. Strength diminishes significantly with rising temperature, albeit the extent of deterioration is contingent upon the specific combination design. The best performing models are subsequently implemented in a Python-based graphical user interface for expedited prediction. Our findings offer a precise and comprehensible methodology for evaluating temperature-dependent compressive strength and facilitate the practical, data-driven design of fiber-reinforced high-performance concrete.

0.26

Analysis of Binary O-type Star Systems Using Recurrent Neural Networks

cs.AI astro-ph.IMastro-ph.SR ·2026-10-16 ·Miguel Flores R., Luis J. Corral, Celia R. Fierro-Santillán, Silvana G. Navarro
unusual
Abstract · 16 pages, 16 figures, 34 images

This work presents an initial analysis of the estimation of effective temperature and surface gravity for simulated O-type binary systems using a recurrent neural network. The proposed system uses 34 equivalent-width measurements from the optical region of a stellar spectrum. Two methods were applied to observed IACOB spectra before normalization: the addition of a blackbody continuum and the addition of a second observed spectrum. The first method produces only small differences in the normalized line profiles and in the estimated physical parameters over 3700-7000~Å. In the second method, the secondary spectrum is added with contributions of 5, 10, 20, 30, 50, and 75 per cent, as defined relative to the original primary spectrum. The general comparison includes 192 simulated systems and 1152 modified spectra. For most of these spectra, the estimated effective temperature and surface gravity decrease as the contribution of the secondary component increases. Individual examples using HD14947, HD190429, HD48099, and HD206267 show that this behavior is not the same for every combination: some estimated effective temperatures fall below the estimates for both isolated spectra, and some surface gravity estimates increase. Comparisons with previously published values also depend on the physical parameter and reference considered. The results show that the proposed system is sensitive to the addition of a secondary spectrum, but do not establish a unique contribution for the secondary component of an observed binary.

0.26

Forecasting Voltage Time Series under Anomalies Using a WGAN Model With Lipschitz Control and Progressive Kernel Critic Strategy

cs.LG ·2026-10-16 ·Najmieh Sadat Safarabadi
benchmark
Abstract

In time-series forecasting tasks, the use of recursive forecasting on high-dimensional data substantially increases the complexity of the prediction process. The main challenge is that long-term dynamics becomes harder to model, since the sequences show greater temporal variability. To counteract this limitation and improve the modeling of long-term dynamics, and more importantly, address high-dimensional recursive forecasting, this paper uses a Wasserstein GAN equipped with a progressive Lipschitz penalty and an adaptive kernel. This model is capable of taking a window of the data stream and forecasting the next values, subsequently reconstructing the time-series stream and dynamically adapting its kernel to each new window. The proposed model is a deep learning framework based on a Wasserstein GAN (WGAN-CLP) with an adaptive kernel to model alterations in streaming voltage time series. To establish the methodological foundation, we review the WGAN-LP formulation another variant model in this class and extend its framework through an adaptive kernel critic model. We then evaluate this extended model using power ratios, correlation coefficient (CC), and relative root mean squared error (RRMSE). The experimental results demonstrate that the proposed model outperforms benchmark methods in both accuracy and robustness. The evaluation shows promising performance of the WGAN-CLP with an adaptive kernel framework in forecasting sequence alterations across the voltage and energy time-series datasets.

0.26

Distillation, Self-Training, and the Fragility of Learned Chemical Reasoning

cs.AI ·2026-10-16 ·Anagha Aneesh, Sebastian Ibarraran, Frank Hu, Jason Liu, Grant Rotskoff
Abstract · 10 pages, 7 figures

Automating synthesis planning for small-molecule drug design requires chemical reasoners that are accurate as well as open to inspection and adaptation, yet frontier language models remain unreliable at predicting reaction outcomes and difficult to deploy, while small open-weight models lack the necessary domain reasoning. We show that frontier supervision can serve as a one-time bootstrap for open models on forward reaction prediction. Our models, Resonance, distill an open-weight teacher's reasoning traces and then learn from their own verified outputs, turning a small base model with near-zero accuracy into a capable reaction predictor, while a stronger student surpasses its teacher through distillation alone. Accuracy, however, does not bring robust reasoning: the small student inherits its teacher's single-pass style, self-training mainly sharpens answers it could already reach, and it readily abandons correct answers under pushback, suggesting that chemical reasoning models should be evaluated on the robustness of their reasoning as well as on the accuracy of their answers.

0.26

SETAREH Transform: A Sorting-Assisted Haar-Wavelet Sparsification Method for Approximate Inner-Product Computation

cs.LG ·2026-10-16 ·Seyed Kamaledin Setarehdan
Abstract · 22 pages, 2 figures, 5 Tables, 21 References

Modern deep learning systems increasingly rely on high-dimensional vector operations, in which large input and parameter vectors are repeatedly stored, transferred, and processed. Although substantial research has focused on reducing neural-network model size through pruning, quantization, knowledge distillation, and efficient architectures, the storage, memory-traffic, and communication costs associated with repeatedly presented high-dimensional input vectors remain important practical considerations, particularly in distributed and edge-computing environments. This paper proposes the SETAREH (Sorting-assisted Energy-concentrating Transform for Approximate Representation using Haar wavelets) transform, a Haar-wavelet-based sparsification approach for approximating the inner product between a fixed high-dimensional coefficient vector and successive input vectors. By applying the same sequence of permutations and Haar transforms to both vectors, SETAREH preserves the exact inner product before truncation. It then retains only a subset of the transformed coefficient-input pairs, reducing their representation cost while introducing a controllable approximation error. To evaluate SETAREH, we consider two complementary experimental settings: a synthetic Gaussian-vector baseline and a real apparent-age estimation task using a convolutional neural network on the APPA-REAL dataset. The results show that SETAREH can approximate the evaluated inner products using substantially fewer transformed coefficient-input pairs than the original vector dimension. The findings support its use as a potential structured data-representation method for repeated inner products with fixed weights, while the practical benefit depends on approximation accuracy and the cost of storing or transmitting permutation metadata.

0.26

What Accuracy Hides In Learned Transformer Programs

cs.LG ·2026-10-16 ·Jonathan Lee, Narmeen Oozeer, Philip Quirke
Abstract

Transformer Programs (TP) (Friedman et al., 2023) constrain a Transformer's architecture so that its learned computation can be extracted as readable Python code. TP expose learned computations, but token accuracy can hide incorrect complete answers. In this work, we examine how program inspection provides insights beyond accuracy scores and helps guide training. We first evaluated TP on nine algorithmic tasks. We compared token accuracy with accuracy on complete output sequences and checked for failures missed by sampled tests. Further, we assessed training data improvements informed by program inspection and failure analysis, finding gains in sequence accuracy on several tasks. We examined a counting task where a program achieved 100% accuracy on sampled test sequences but failed on another input using the symbols and lengths allowed during training. The program counted correctly, but the required answer was missing from its output labels. After adding training examples with larger counts, the retrained models passed exhaustive evaluation over the bounded domain. In the frequency ranking task, inspection showed how correct counts contributed to incorrect answers in programs trained before and after changes to the training data. Finally, we studied constraints imposed by TP training and operations. The TP training process can lead to accuracy losses late in training. Our reference constructions also show how composing the restricted TP operations can use many layers or large lookup tables.

0.26

Learning Administrative Boundary-Aware Location Encoders via Reverse Geocoding

cs.LG cs.AI ·2026-10-16 ·Lakshay Sharma, Alex Marin
Abstract

Location encoders learn compact representations of latitude--longitude coordinates for downstream geospatial prediction tasks. Existing approaches are pretrained using continuous or naturally occurring supervision, such as imagery, climate variables, or species distributions, and therefore do not explicitly capture administrative boundaries or place names, despite their strong influence on many real-world geospatial phenomena. We introduce, to our knowledge, the first method for learning administrative boundary-aware location encoders, using reverse-geocoded text as a self-supervised training signal. Our approach can be applied to existing location encoder architectures and consistently improves their downstream performance. We further introduce RevGeoRQ, a custom architecture tailored to reverse-geocoding supervision, which yields additional gains and exceeds contemporary location encoders on several downstream tasks. Finally, the use of human-interpretable reverse-geocoded strings and residual-quantized semantic IDs provides a more interpretable representation of geographic location than existing approaches. Code and model weights to be released soon.

0.25

Self-Reflective Chain-of-Thought (SRCoT): Evidence-Based Text Generation via a Closed-Loop Retrieve--Reflect--Revise Pipeline on Edge LLMs

cs.AI cs.CL ·2026-10-16 ·Fabian Broszio, Michael Hoffmann, Adrian Paschke
Abstract · 28 pages, 2 figures. Extended technical account; a related brief report based on overlapping work has been submitted to Frontiers in Disaste…

Large language models (LLMs) can generate fluent outputs that lack source support or conflict with local operational constraints. We introduce Self-Reflective Chain-of-Thought (SRCoT), a modular architecture that organizes generation as a closed Retrieve--Reflect--Revise loop. SRCoT separates environment state, persistent knowledge, accumulated experience, and short-term memory into provenance-labeled stores. BM25 retrieval supplies local context; subsequent passes expose intermediate drafts, missing-information markers, follow-up queries, and optional claim--evidence mappings. These artifacts inform retrieval, revision, clarification, or abstention decisions. We examine the architecture through qualitative, trace-based case studies in two fully synthetic settings: emergency medical services and Munich-Sim, an emergency-operations simulation. Local prototype runs on an Apple M4-family laptop with 128 GB memory illustrate source-labeled documentation, explicit gap marking, and structured dispatch-style outputs. The traces also reveal provenance-assignment errors, incomplete evidence coverage, and unsafe assumptions that remain possible despite prompting constraints. These cases demonstrate report-level inspectability and local execution, not comparative superiority, hallucination reduction, clinical validity, or operational safety. The architecture separates retrieval and control from the model interface, but behavior across different models remains to be evaluated. SRCoT is a research proof of concept requiring human review, with its principal contribution lying in an inspectable control procedure rather than a new model.

0.25

DrugHarness: An Evolvable Agent Harness for Computational Drug Discovery

cs.AI ·2026-10-16 ·Yuxing Chen, Zhisheng Zhang, Cheng Tang, Wei Li, Minghui Zhang +18 more
hot direction
Abstract

Computational drug discovery requires agents to compose scientific operations and adapt to changing evidence and constraints. Here we introduce DrugHarness, an evolvable agent harness coupling composable scientific execution with conditional skill induction. Its environment integrates over 80 scientific operations, biomedical resources and task-specific code, preserving associations among candidates, results and artifacts. From execution trajectories and outcome feedback, DrugHarness induces action hypotheses with observable applicability conditions, recommended actions and supporting evidence. An independent review agent evaluates candidate skills derived from these hypotheses through source verification and, when needed, targeted execution trials. Admitted skills guide subsequent tasks and are revised through feedback. Evaluation on MolBench and SMDD-Bench demonstrates capabilities in molecular screening, optimization and constrained design. Without evolution, DrugHarness exceeds reported MolClaw-CC results by 10.8 percentage points in affinity-comparison accuracy and 55% in mean top-three active hits on MolBench. Retrospective evaluation on skill-development tasks shows that evolution increases Qwen's affinity-comparison accuracy from 62.2% to 81.1% and optimization success from 79.5% to 87.2%. In a prospective HPK1 study, DrugHarness achieves a more favorable median docking score than Biomni among system-specific candidates. A synthesized compound achieves a cellular IC50 of 14.19 nM, a 2.64-fold potency improvement over the best input reference, with activity competitive with two selected compounds designed by experienced medicinal chemists. These results support conditional skill evolution for scientific agents and demonstrate a workflow connecting computational drug design with experimental validation.

0.25

OwlSight: A Panoramic Monitor for Chain-of-Thought Reasoning

cs.AI ·2026-10-16 ·Lewen Yan, Yanxu Zhu, Jie Zhang, Jing Shao
benchmark
Abstract · We wish to withdraw this preprint because several fundamental errors have been identified in the manuscript

Chain-of-Thought (CoT) monitoring offers actionable evidence for oversight AI systems, with the potential to enable early detection and intervention. However, existing approaches are focused on specific tasks, whereas the regulation of AI, particularly reasoning models, requires monitoring a broader range of tasks. We introduce Factor for CoT monitoring, a strategy that supports customizing monitoring requirements in the prompt, which enables a single monitor to cover different tasks by relying on different factors. Furthermore, we introduce two types of factors, fine-grained and coarse-grained (e.g., monitoring whether a specific email address has been leaked versus monitoring privacy leakage), to accommodate diverse user monitoring needs. Based on this, we develop a CoT monitor named OwlSight that relies solely on factor to monitor behaviors related to multiple dimensions, including capability, safety, and trustworthiness. Experiments across 15,971 reasoning trajectories show that OwlSight surpasses general-purpose LLMs by 1.32-30.57 and 11.71-27.81 points at fine and coarse granularity, respectively, and achieves state-of-the-art results on OpenAI's CoT monitorability benchmark. It also generalizes to unseen models' CoTs and remains robust to Factor rephrasings, CoT compression, and obfuscation, offering a promising paradigm for CoT monitoring that can contribute to the oversight of frontier AI.

0.25

Not All Test-Time Compute Is Evidence: Model-Specific Bias in LLM Reasoning

cs.AI ·2026-10-16 ·Ming Shen, Tianjun Yao
Abstract

Test-time scaling methods implicitly assume that additional computation from the same model provides independent evidence. We show that this assumption fails within a single reasoning model: errors are correlated across repeated samples, across reasoning depth, and in the model's own self-evaluation, so more compute from the same model is not more evidence. We introduce DIA (Disagreement-triggered Independent Arbitration), which treats disagreement between an intermediate-stage and the final-stage prediction as a per-question signal of reasoning instability and resolves it by asking a small panel of heterogeneous models to solve the question independently, without seeing either candidate, and aggregating their answers directly. Across seven benchmarks and two base models, DIA outperforms full chain-of-thought in 12 of 14 settings, including a 7.00% gain on MuSiQue, while activating on only $\sim 17\%$ of examples and using up to 47.2% fewer tokens than a multi-agent baseline. Ablations show that both independent solving and solver heterogeneity are necessary for these gains.

0.25

Beyond Large Language Models: A Mental World Model Framework for Mental Modeling, Simulation, Evolution, and Interpretability

cs.AI ·2026-10-16 ·Qingshi Wang, Huiqi Tong
Abstract · 12 pages, 10 figures

Understanding and simulating how subjective cognition and mental states emerge, evolve, and shape behavior is essential for advancing artificial intelligence toward deeper social cognition and general intelligence. Existing large language models (LLMs) and Theory of Mind (ToM) methods primarily infer mental states from linguistic information, lacking unified mechanisms for subjective world construction, cognitive-psychological interactions, dynamic evolution, and historical dependencies. This paper introduces the Mental World Model (MWM) and its implementation, the Multi-Branch Mental Evolution Architecture (MMEA), which explicitly constructs subjective worlds, models interactions between cognition and mental states, and simulates continuous mental evolution shaped by historical experiences. On DynToM and RecToM, MMEA achieves 96.48% and 94.00% accuracy, respectively, outperforming leading LLMs and specialized ToM methods. Additional experiments demonstrate improvements of 28.7, 39.3, and 16.6 percentage points in mental state and behavior prediction, long-sequence prediction, and implicit mental state inference, respectively. The results demonstrate that MMEA provides an interpretable framework for continuous mental modeling, enabling AI to move beyond surface-level semantic representations to infer individuals' underlying mental states from linguistic information and dynamically simulate their internal cognitive processes and evolution over time. MWM also provides a promising foundation for causal reasoning, personalized AI, sustained human-AI interaction, and future advances toward artificial general intelligence (AGI).

0.25

Where Research Ideation Begins: Forward-Citation Supervision for Actionable Limitation Generation

cs.AI ·2026-10-16 ·Liyang Zhao, Yong Cao, Haibin Ling
Abstract · 19 pages, 4 figures, including appendices

Automated research ideation has advanced rapidly, with multi-agent workflows that generate whole research ideas end to end. These frameworks leave largely unexamined the step that seeds a good idea: identifying an actionable limitation from literature. In this paper, we focus on this first step in the research ideation pipeline. Starting from a pilot study, we find that as large language models (LLMs) scale from 0.6B to 32B parameters, they become far better at recognizing which limitation later work addressed, but no better at generating one, suggesting that the bottleneck lies in the supervision signal rather than model capacity. We therefore construct ActLim-Bench from forward citations, yielding 18,279 evidence-linked supervision triples with human verification. On this signal, we propose a three-stage post-training pipeline: supervised fine-tuning, verifier-filtered rejection-sampling fine-tuning, and reinforcement learning with the verifier as an online dense reward. Experimental results show that our method achieves the highest recall score compared to existing methods among all frontier models. We further show that our generated limitations are preferred in 64.9% of non-tied human comparisons and improve downstream ideation systems. We will release our data and code upon publication.

0.25

Multiple Turns, Short Horizons: Measuring Effective-Horizon Compression in Jailbreaks

cs.AI ·2026-10-16 ·Zhipeng Wei, Michael Lee, N. Benjamin Erichson
Abstract

Multi-turn jailbreaks use feedback from a victim large language model to adapt their queries over several turns. However, once an effective trajectory has been found, it is unclear how much of the earlier query history is still needed. In this work, we study this by removing early attacker queries and running the remaining queries again in a new conversation while regenerating the victim's responses. We define the effective horizon as the shortest remaining query sequence that recovers the same attack progress. Empirically, we find that roughly half of the original query sequence is sufficient on average to recover the same attack progress. We call this phenomenon effective-horizon compression. We show that this shorter history can recover the best point reached during an unsuccessful attack and provide a better starting point for continuing the attack. It also helps identify which earlier queries should receive less learning signal during attacker training, and downweighting those queries improves performance. We further find that short reward-propagation windows preserve calibrated defense performance close to full-trajectory returns. These results show that multi-turn attacks can contain more queries than are needed to recover the progress reached later in the interaction.

0.25

Boundary Detection for High-Dimensional Point Clouds

cs.LG cs.CGstat.ML ·2026-10-16 ·Abigail Hickok, Valerio Galanti, Osama Al Dalahmah, Andrew J. Blumberg
Abstract · 31 pages; 23 figures

We introduce two new methods for boundary detection in high-dimensional point cloud data, intended to work in situations with non-uniform density, high ambient dimension, and substantial noise. Our first method fits a separating hyperplane on local neighborhoods after doing local dimension reduction---this method works best when one can reliably estimate the intrinsic dimension of the data. Our second method measures the directional uniformity of local neighborhoods, and has the best overall performance on synthetic data. In addition to pointwise numerical boundary scores, our algorithms yield an estimated normal vector at each boundary point. We validate our methodology on synthetic data as well as two real examples: we analyze spatial transcriptomics data, where we show that we can detect the boundary of a tumor, and we also study LLM activation geometry.

0.25

CIIR - Class-Indexed Interface Representation Learning

cs.AI cs.LG ·2026-10-16 ·Doo Yon Kim, Christian Wallraven
Abstract · 17 pages, 6 figures

Multimodal learning commonly assumes paired or synchronized observations, yet independently collected datasets rarely provide such correspondence. We introduce CIIR, a modular encoder--decoder framework that connects heterogeneous datasets through a class-indexed address space while preserving each dataset's native input format and output module. A known local index selects address $k$, and an input-dependent 128-dimensional address representation Z_k(x) is written to that address. To demonstrate CIIR, we learn this address space from handwritten digits, clothing images, spoken-digit spectrograms, and hand signs without synchronized or instance-matched samples. Newly trained label-free receivers recover the requested addresses with 94.7% macro-average Address Realization Accuracy. We then attach the unseen SVHN dataset after freezing the base address space and train only lightweight adapters, achieving 90.2% MNIST>SVHN and 82.3% SVHN>MNIST end-to-end Address Transfer Accuracy without target-side decoder labels. To demonstrate generalizability of CIIR, we reuse the ten addresses as code elements in collision-free arithmetic and hierarchical address codes for all 100 CIFAR-100 classes and evaluate direct arithmetic-code realization through the existing receivers. These results establish CIIR as a compact interface for modular cross-dataset communication and post-hoc dataset attachment without exhaustive paired multimodal supervision.

0.25

LocalIntent: Outcome-Specific Structured Generation for Backend-Compatible Smart Building Automation with Local LLMs

cs.AI ·2026-10-16 ·Tianyang Liu, Jiayue Wang, Zhanhao Cheng, Changyang Wang, Shaoshu Zhu +2 more
Abstract

Locally deployed large language models (LLMs) can reduce reliance on remote cloud services for natural language smart building automation, but backend-facing processing requires correct outcome selection, alignment with local capabilities, and valid structured outputs. We investigate whether a complete outcome-specific prompting design improves compact-model outputs beyond generic route-then-generate prompting. The proposed LocalIntent approach routes requests to ordinary or persistent-rule branches: the former produces commands, clarifications, or rejections, while the latter generates trigger-action rules, followed by deterministic validation. We evaluate four locally executed Gemma and Qwen models spanning 2B-9B parameters on 101 non-persistent and 54 persistent-rule records, comparing Monolithic prompting, generic two-stage prompting, LocalIntent, and validation-guided repair. Across the evaluated datasets, LocalIntent achieves higher intent understanding, device identification, and action/parameter correctness than generic two-stage prompting, while clarification/mode correctness remains model-dependent. The largest gains occur for rules, where LocalIntent improves intent understanding from 26.4% to 80.6%, device identification from 59.3% to 84.7%, rule-mode correctness from 49.1% to 98.6%, and action/parameter correctness from 5.6% to 44.9%. For non-persistent requests, pooled gains are smaller: intent understanding increases from 39.6% to 53.2%, and action/parameter correctness from 30.7% to 42.8%.

0.25

Where Latents Belong and How They Flow: Joint Identification of Structural Roles

cs.LG ·2026-10-16 ·Xintong Li, Kun Zhang, Yujia Zheng
Abstract

Latent factors in dynamical systems can play different roles across observations and time. For example, an idea in a group meeting may be expressed by only some participants, inherit information from earlier discussions, and shape what is said next. We formalize this notion as the structural role of a latent factor, describing where it is expressed, whether it receives influence from the past, and whether it transmits influence to the future. The main challenge is correspondence, since the three structures may be identified under different permutations, while a structural role is defined only when they are aligned to the same latent coordinate. We show that, under suitable structural and variability conditions, all three roles are jointly identifiable under a common permutation at each time point. Guided by this theory, we develop RoleTrace, a framework that traces structural roles in a shared latent coordinate system. Synthetic experiments validate structural recovery, while experiments on multiparty meetings examine the identified roles in real sequential interactions.

0.25

QuAD: Quantitative Assessment via Distribution-Alignment for Qualitative Data Quality Enabled by a Content-Aware Multi-Agent System

cs.AI cs.HC ·2026-10-16 ·Wenshuo Zhang, Leixian Shen, Qiqi Zhou, Shuchang Xu, Feiyang Hao +4 more
Abstract

Assessing whether qualitative responses collected from semi-structured questions are ready for a particular analysis goal is a persistent challenge. Manual qualitative analysis is interpretive and labor-intensive, while existing automated methods often lack a dataset-level perspective for deciding whether a response set supports a researcher-selected analytical lens. We introduce QuAD, a framework for quantitative pre-analysis assessment of qualitative response datasets. QuAD compares a reconstructed intrinsic cluster distribution reconstructed from responses with an extrinsic, researcher-defined analytical lens derived from metadata. QuAD is structured into a four-quadrant assessment whose metrics are read jointly, from lens-level relations to within-cluster variation, to provide fitness-for-purpose signals for revealing general patterns and unique insights. Because this assessment relies on lens-conditioned inter-response dissimilarity, we propose a content-aware multi-agent system with two-stage instruction generation and adaptive anchors to improve scoring consistency. We validate QuAD through controlled experiments on synthetic data, including robustness and baseline comparisons, and a real-world social survey case study. The case study shows how QuAD quantitatively assesses whether a selected analysis axis is promising and also surfaces representative or unusual responses for human interpretation.

0.25

Muon Operates Beyond the Edge of Stability

cs.LG ·2026-10-16 ·Niko Covic, Baraq Lipshitz, Alexander Padula, Stefan Visnjic
code/data
Abstract

We study the Muon optimizer using sharpness and provide the first empirical evidence that the Flat Minima Hypothesis also applies to Muon. Further, we derive Muon's implicit preconditioner to analyze its potential Edge of Stability behavior. Experiments on CIFAR-10 show sharpness strongly correlates with generalization gap within each optimizer (Pearson $r = 0.45$-$0.69$, $p \leq 10^{-4}$), confirming the hypothesis holds for Muon similarly to Adam and SGD. However, Muon operates above the classical sharpness-based Adaptive Edge of Stability threshold that bounds diagonally-preconditioned optimizers like Adam and RMSprop, yet remains stable. Our findings suggest Muon operates in a fundamentally different regime than diagonally-adaptive methods. Defining the exact stability boundary for Muon remains an open question. All code is available at https://github.com/NikoCovic/DL-project.

0.24

The Loss or the Architecture? What Matters in Survival Analysis with Competing Risks

cs.LG stat.ML ·2026-10-16 ·Julie Alberge, Gaël Varoquaux, Judith Abécassis
benchmark
Abstract

Often, \emph{eg} in healthcare and predictive maintenance, what matters is which of several competing causes will strike first, and with what probability: survival analysis with competing risks. Corresponding deep learning methods are developed as bundles: a new architecture together with a new training loss, tested on one to five datasets against a handful of baselines. Thus, a state-of-the-art result cannot be attributed to the architecture, the loss, or the evaluation protocol. The coupling is not accidental: with one incidence curve per cause, a network's head is tied to the losses expressible on it. We vary the two axes independently, three losses, two of them strictly proper, on four backbones, across two modalities that tease the axes apart: 20 tabular and 3 pathology-image datasets. One loss wins in both: modelling cumulative incidence directly under censoring weighting. It brings twice as much as benefits a good choice of backbone on competing-risks data, and no more than it on single-event data: the loss is decisive for competing risks. The backbone matters too, and must match the modality: tabular networks lead on tables, pretrained vision encoders on images. The two choices compound rather than substitute. Proper scoring rules reveal these dynamics; the concordance index (C-index), which is not proper, does not separate models: neither axis moves it, not even tuning on the loss built to improve it. This is an avenue of progress: composed, the two give a model nobody has tried, RealMLP or TabM trained with the cumulative-incidence loss, which takes the best Brier score in our benchmark.

0.24

Wavelength Diversity Expands the Expressive Power of Optical Computing

cs.NE physics.opticsphysics.app-ph ·2026-10-16 ·Shiqi Chen, Che-Yung Shen, Parnian Ghapandar Kashani, Yongkang Cheng, Aydogan Ozcan
Abstract · 34 Pages, 10 Figures

Universal function approximation underpins neural networks' capacity to represent nonlinear relationships, but realizing this capability in optical hardware requires engineering the computational features encoded in light. Here, we show that wavelength diversity enhances optical function-approximation expressivity through wavelength-scaled Fourier features generated by a shared phase encoder, without increasing the number of input-encoded harmonics. Multiple wavelengths contribute to the representation of each output function rather than merely serving as independent computational channels. We establish theoretical scaling laws and approximation-error bounds, revealing that, for nondispersive encoders, the spectral endpoints determine the accessible computational bandwidth, while intermediate illumination wavelengths enrich the available harmonics within that bandwidth. Numerical studies demonstrate improved approximation of real- and complex-valued nonlinear functions, with a representative design achieving more than a 100-fold reduction in mean-squared error relative to tested monochromatic optical processors with the same number of trainable phase features. Experimentally, we validate enhanced approximation accuracy and demonstrate the simultaneous approximation of 4,136 distinct nonlinear functions using a compact, single-layer diffractive processor under five-wavelength illumination. These results establish wavelength diversity as an additional physical resource for scaling the expressive power of optical computing and enhancing the accuracy of parallel nonlinear function approximation.

0.24

Learning Personalized Memory Gating from Delayed One-Sided Feedback

cs.LG ·2026-10-16 ·Kaixuan Ji, Zixiang Chen, Wenting Zhao, Cheng Qian, Heng Wang +7 more
theory
Abstract

Long-term memory enables language-model agents to personalize future interactions, but a finite memory bank cannot preserve every extracted fact. Existing memory gating policies are typically shared across users and therefore fail to account for user-specific patterns of future reuse. In this paper, we study memory gating at the fact level. We first propose a retrieval-based labeling strategy that labels each fact by whether it is later retrieved and incorporated into responses. We then formulate online memory gating as contextual apple tasting with delayed one-sided feedback, and propose Warm-Start UCB, which combines offline learning from existing data with online adaptation to the current user. Our theoretical analysis establishes a high-probability regret upper bound of $\tilde{O}(Tε\wedge d\sqrt{T})$ when the delay window is at most $O(\sqrt{T})$, where $T$ is the total number of rounds, $ε$ is the initialization error, $d$ is the dimension of the feature map, and $\tilde{O}(\cdot)$ hides constant and poly-logarithmic factors. The results of our experiment demonstrate that our method outperforms all baselines under different memory budgets and history lengths, establishing it as a promising personalized memory gating policy.

0.24

Self-Improving Video-to-Action Models

cs.AI cs.CV ·2026-10-16 ·Hyunwoo Ryu, Eric Ming Chen, Max Wilcoxson, Vincent Sitzmann
code/data
Abstract · 33 pages, 18 figures, 6 tables. Project page: https://wilcoxsonm21.github.io/mirror/

Video generative models are an emerging tool for embodied agents: they can serve as scalable visual planners, generating videos that depict how to complete previously unseen tasks. However, executing these plans typically requires a video-to-action or "inverse dynamics" model trained on large, human-collected video-action datasets, severely limiting scalability. We introduce Mirror, a self-supervised approach that learns video-to-action mappings through environment interactions, without prior action data or hand-crafted rewards. Mirror learns by attempting to follow a visual plan by executing its best guess of the underlying actions in the environment. Mirror then trains on the resulting video-action pairs, incrementally self-improving and following the visual planner more closely. Across Atari, OGBench, and DeepMind Control, Mirror outperforms leading learning-from-observation methods with learned video plans, and achieves near-expert performance with oracle plans. On Minecraft BASALT, it outperforms methods with access to expert actions or hand-engineered task knowledge. In Minecraft survival, it is the first method to acquire an iron pickaxe through self-improvement alone, without rewards or expert action labels. In the offline setting, with expert action labels but no environment interaction, it sets a new state of the art in obtaining diamonds. Our results suggest that advances in video models can translate into increasingly capable agents without a corresponding need for hand-crafted rewards or large collections of expert action data.

0.24

Structural Affinity: An Algorithm for Coherent Cage-Based Shape Design

cs.LG cs.GRcs.RO ·2026-10-16 ·Aviraj Newatia, Shengan Yang, Hantao Zhong, Rika Antonova
Abstract · 27 pages, 16 figures

Cage-based deformation (CBD) parametrises 3D geometry through a small set of control handles, effectively enabling a variety of tasks from animation and interactive modelling to tool-based co-design, neural deformation transfer, and aerodynamic shape optimization. Many optimization approaches relying on CBD employ regularisation strategies to reduce the distortion of the deformed mesh by encouraging smooth, regular cage deformations. These techniques offer soft guidance to an optimization algorithm towards desirable deformations, but compete with the task objective rather than changing the deformation parameterization itself. In this work we introduce Structural Affinity for CBD, an explicit regularisation method which enables smooth deformations through control cages by exploiting a two-cage structure to couple cage-handle motion directly. Structural Affinity precomputes an affinity matrix using mean-value coordinates, binding the inner control-cage to a coarser outer-cage. This matrix defines how much a cage handle displacement propagates to neighbouring handles which substantially decreases and often eliminates cage self-intersections, reduces mesh distortion, and adds linear overhead at deformation time. Across robotic control--morphology co-design, source-to-target neural deformation transfer, and aerodynamic wing shape optimization, Structural Affinity produces more regular deformations and substantially reduces reliance on hand-designed structural loss terms. We report consistent gains on three complementary diagnostics: the cotangent Laplacian, a proxy for surface smoothness; cage distortion, capturing whether control handles move coherently; and surface roughness, which measures undesirable oscillation in wing geometry.

0.24

Expressivity of IQP Feature-Map Quantum Kernels via Reproducing Kernel Hilbert Spaces

cs.LG quant-ph ·2026-10-16 ·Valeria Repetto, Stefano Melchionna, Romina D'Aurizio
Abstract · 19 pages, 11 Figures , Supplementary

Quantum kernels are among the most promising methods in quantum machine learning, providing a natural way to embed data into high-dimensional Hilbert spaces via quantum circuits and capturing non-linear relationships in data. Their learning behaviour is governed by the function class they represent, which is formally characterized by the structure of the associated Reproducing Kernel Hilbert Space (RKHS). However, a theoretical characterization of this function class, and of how it depends on the choice of feature map and its hyperparameters, is still lacking. In this work, we focus on the Instantaneous Quantum Polynomial (IQP) feature map and derive a Fourier representation of the associated RKHS generators as finite sums of trigonometric functions evaluated on polynomially transformed input variables, with a finite set of accessible frequencies determined by the circuit construction and scaled by the bandwidth hyperparameter. This representation identifies three bandwidth regimes with qualitatively distinct expressive properties from localised, through multimodal, to periodic. Numerical experiments on synthetic datasets with controlled periodic and multimodal structure, across data dimensions and feature-map configurations, reproduce the three regimes in both supervised classification and unsupervised anomaly detection, benchmarked against a classical Gaussian kernel. These results make the relationship between expressivity and data structure explicit for IQP kernels: expressive power translates into a practical benefit only when the accessible frequencies match with the structure of the data, which provides a criterion for selecting the bandwidth when the periodic or multimodal structure of data is known.

0.24

A general recipe for data-adaptive anytime-valid envelopes for iterative algorithms

cs.LG stat.MLmath.OCstat.ME ·2026-10-16 ·Soham Bonnerjee, Aaditya Ramdas
Abstract · 90 pages, 4 figures

Most convergence guarantees for iterative stochastic algorithms are stated at a fixed iteration count and do not justify stopping at a data-dependent time. Recent anytime-valid envelopes do, but deterministic ones are typically too conservative for practical stopping, and existing data-adaptive ones are derived case-by-case, each concerned with a specific algorithm and a specific noise model and boundary construction. This paper provides a general recipe for sharp, data-adaptive anytime-valid envelopes for a large class of iterative algorithms. Our approach has two key steps: an optimization module derives a one-step recursion that performance measures of many such algorithms satisfy, reducing the error analysis to time-uniform concentration for predictably weighted sums of noise. Then, a statistical module ports the sub-$ψ$ condition to vector-valued processes, uniformly over predictable weights, and derives off-the-shelf boundaries via mixtures, stitching, and product betting martingales. Combining the two yields readymade envelopes for many iterative algoithms including but not limited to stochastic gradient descent (SGD), Oja's algorithm, temporal difference learning, Q-learning, and Bures-Wasserstein SGD, many with no prior data-adaptive anytime-valid guarantees. Specifically, for SGD, our framework recovers, often under weaker conditions, or significantly sharpens every existing envelope we are aware of. Our constructions come with explicit implementation details and are substantiated by numerical experiments.

0.24

Data-Driven Resource Aggregation for Scalable Digital Twin and Simulation Model Generation

cs.LG ·2026-10-16 ·Yucheng Wu, Alp Akcay
Abstract

As industrial digitization accelerates, the use of digital twins has become widespread. However, modern manufacturing systems are highly dynamic and frequently reconfigured, making manual model development and updates increasingly difficult. Recent research on automated simulation model generation shows that historical data such as event logs can yield simulation models automatically. However, modeling every resource at full granularity leads to unnecessary complexity and computational cost. We propose a method for automatically selecting and aggregating resources to simplify simulation modeling. Our method uses ideas from inverse reinforcement learning to learn behavioral value representations of resources from production event logs. This process identifies and aggregates resources with similar functional contributions. Experiments on manufacturing systems, using throughput and system time as performance measures, demonstrate that value-based resource aggregation performs well when the goal is to substantially reduce complexity while preserving performance accuracy.

0.24

A Minimal ReLU Classifier on Non-Linearly Separable Data: Convergence and Implicit Bias at and beyond the Edge of Stability

cs.LG ·2026-10-16 ·Quan Nguyen, Robin Yadav, Param Mody, Mark Schmidt
theory
Abstract

Gradient descent (GD) with large learning rates can successfully optimize neural networks even though classical smoothness-based learning rate conditions are violated. Yet theory for this regime largely covers linear models or linearly separable data. We analyze a minimal setting outside both regimes. Specifically, a two-neuron ReLU network trained with logistic loss on 1D non-linearly separable data. The objective is nonconvex and nonsmooth, and GD exhibits edge-of-stability (EoS) behavior. We identify an active cone in parameter space that contains every perfect classifier, where the problem becomes piecewise logistic regression with a varying effective dataset. Within this cone, we prove a finite-time loss bound for constant learning rates and show that GD converges in direction to an explicit max-margin vector, while the orthogonal component converges to the unique minimizer of a non-strongly convex potential. We further establish catapult results where exponentially small loss is reached after a single loss spike. Furthermore, we propose an adaptive schedule that lets learning rates grow to infinity without leaving the active cone. We prove that this schedule achieves a loss rate of $O(1/(T\log{T})$, which is faster than the standard $O(1/T)$ rate. Our theoretical results extend to local convergence on more general network and are validated on MNIST dataset. Overall, we show that this setup provides a tractable neural-network model for studying large and adaptive step sizes beyond linear predictors and separable data.

0.24

Acquisition Function-Guided Generation for Bayesian Optimization over Discrete Spaces

cs.LG ·2026-10-16 ·Tommaso Ancilli, Thomas Wedenig, Henry Moss, Robert Peharz
code/data
Abstract

Bayesian optimization (BO) over discrete spaces is challenging due to missing gradients and the sparsity of valid solutions. Latent-space BO sidesteps both issues by fitting a surrogate in the latent space of a generative model, which however substantially complicates the direct incorporation of domain knowledge that is naturally expressed in the original discrete space. In this work, we instead combine a pretrained generative model with a surrogate defined directly in data space, where inductive biases such as domain-specific kernels can be readily incorporated. Rather than maximizing the acquisition function over the discrete design space, our method samples candidates from the generative model under a distribution tilted by the acquisition function. To recover useful gradients for sampling, we embed this tilted distribution into a diffusion bridge that yields a smooth, posterior-averaged acquisition function. This leads to \emph{acquisition function guided generation} (AFGG), which we instantiate with a categorical diffusion model and an autoregressive normalizing flow. Across a broad suite of molecular benchmarks, the diffusion model variant performs on par with leading BO methods, while the flow model variant outperforms state-of-the-art methods in most cases and remains highly competitive otherwise. Our implementation is available at https://anonymous.4open.science/r/afgg.

0.24

Margin Certificates for Voting Classifiers under Mixing Dependence

cs.LG math.STstat.ML ·2026-10-16 ·Adria Binte Habib, Ibne Farabi Shihab
theory
Abstract

Margin bounds explain why weighted votes generalize after training error vanishes, but dependent observations require concentration arguments that keep the dependence model and failure budget explicit. We study stationary mixing streams without treating dependence as a formal substitution of an "effective sample size." For a finite base class, we combine independent-block Rademacher theory for stationary $β$-mixing processes with a clipped margin loss to obtain an explicit uniform voting-margin certificate, whose dependence penalty is set by the number of approximately independent blocks and the coupling budget $4(μ-1)β(a)$. A black-box transfer theorem shows that any i.i.d. voting-margin guarantee remains valid on a regularly spaced subsequence, with only an additive $β$-mixing failure probability. For votes fixed independently of a stationary finite-state reversible Markov stream, spectral concentration gives the cleaner $\sqrt{t_{rel}/n}$ rate, and a sticky-refresh construction shows this exponent is unavoidable. The constant is statistic-specific: the effective sample size is $n/τ_{int}(g)$, with $τ_{int}(g) \le 2t_{rel}-1$. When the dependence scale is unknown, a batch-means Student-t interval is valid up to an explicit remainder vanishing in the number of batches, with width governed by $τ_{int}(g)$ rather than the worst-case $t_{rel}$. Experiments show: the learned-vote certificate is nonvacuous once the block count is large; a two-timescale construction gives a batch-means interval 8.1x narrower than the worst-case interval at nominal coverage (0.945); block-bootstrap and batch-means intervals undercover under strong dependence while the certificate stays valid; and on real, ordered UCI HAR sensor data the dependence-agnostic certificate reaches 0.95 coverage where naive intervals reach only 0.38.

0.23

Agents' Last Doubt: Knowing When to Act and When to Step Back in Tool-Using Agents

cs.AI ·2026-10-16 ·Qisai Liu, Qingyue Jiao, Bangwei Guo, Wenxi Dai, Changyi Yang +5 more
dataset
Abstract

Tool-using LLM agents repeatedly face decision points under uncertainty, choosing between autonomous action, additional information gathering, and human escalation. These decisions require balancing task success against interaction cost and the risk of irreversible errors. Learning to make these trade-offs is challenging because existing interaction trajectories record only the outcomes of actions taken, leaving the consequences of alternative actions unobserved. To address this gap, we construct a counterfactual dataset containing 3,115 action outcomes from 1,036 decision points in ToolSandbox. At each decision point, we execute each legal action and evaluate its downstream outcomes based on task completion, interaction cost, and irreversible errors. Using this dataset, we train SURE-Act, a controller that selects actions for a frozen backbone agent at inference time. Our experiments show that learning to rank alternative actions improves offline action-selection utility by 0.0309 over a regression-only baseline. We further show that adding SURE-Act enables end-to-end gains of up to 0.0575 in ToolSandbox's Overall Score with the Qwen3-4B agent. Moreover, SURE-Act outperforms adapted decision rules from SAGE-Agent and Calibrate-Then-Act by 0.1506 and 0.1438, respectively. This work establishes the first counterfactual dataset for ToolSandbox and provides actionable insights into decision-making under uncertainty for tool-using agents.

0.23

Characterising and Analysing Retrieval-Augmented Generation Techniques

cs.AI cs.LG ·2026-10-16 ·Emmanuel Igbekele, Olawande Daramola
Abstract

The advent of Large Language Models (LLMs) across many application domains has increased the need for Retrieval-Augmented Generation (RAG) to reduce hallucinations and improve LLM performance. Thus far, several RAG techniques with impressive features and capabilities have been proposed in the literature, making it a herculean task to select the most appropriate RAG techniques for specific problem scenarios. Thus, we propose a conceptual framework comprising main dimensions and sub-dimensions for characterising (defining "what it is") and analysing (defining "how well it does") existing RAG techniques. The conceptual framework was developed through a qualitative analysis of the attributes of RAG techniques reported in the literature. It can serve as a pre-empirical filtering mechanism for researchers and practitioners in selecting the highest-fit RAG candidates that meet specific user and task requirements before empirical benchmarking of RAG techniques, which can be computationally and financially expensive.

0.23

Cortex: A Fixed-Point Theory of Governed Coding Agents - Technical Report

cs.AI cs.LG ·2026-10-16 ·Marius-Constantin Dinu, Florian Zeba
benchmark
Abstract · 27 pages, 8 figures, 9 tables

A coding agent combines a large language model (LLM) with a harness that plans, edits, and executes code. We study Cortex, a supervisory layer for pre-execution governance, requirement tracking, and iterative validation and repair. We formalize requirement closure on a finite lattice. For a fixed requirement set and a sound, monotone validation rule with persistent evidence, exhaustive iteration reaches the least fixed point with no more strict increases than there are requirements. Fair consequence scheduling reaches the same closure. A separate stochastic model bounds expected completion time under a uniform positive-progress assumption. These results depend on their stated assumptions; artifact repair, fallible validation, and finite execution budgets do not satisfy them automatically. We define Gauntlet's trajectory-similarity and build-gated scores, attack-success estimators, Wilson intervals, and case-cluster bootstrap summaries. Effective-feedback accounting counts newly certified requirements, rather than estimating their information value or utility. The five evaluation families support raw-versus-governed comparisons. We report retained benchmark observations, including rescoring of existing artifacts; constructed outcomes are excluded. Unequal observation sets and incomplete provenance prevent causal claims about governance or measured long-horizon scaling.

0.23

The Judge Can Debias Itself: Awareness and Self-Debiasing for Reliable LLM-as-a-Judge

cs.AI cs.CL ·2026-10-16 ·Youbing Yin, Jingyu Wu, Anoop Kumar, Scott Novotney, Daben Liu +1 more
top venue
Abstract

LLM-as-a-Judge systems are increasingly employed for automated evaluation, yet remain susceptible to systematic biases that undermine their reliability. We demonstrate that the judge can effectively debias itself without additional training. We propose two lightweight strategies: bias-aware prompting, which explicitly alerts the judge to potential target biases prior to evaluation, and self-debiasing, in which the judge critiques the initial judgment for bias influence. To broaden the evaluation scope, we introduce three novel benchmarks--authority bias (fabricated citations), confidence bias (assertive phrasing), and formatting bias (structured presentation)--and evaluate across seven bias types in total. Across three frontier models, self-debiasing improves performance in 20 out of 21 model-bias combinations with gains up to 15 percentage points, while bias-aware prompting also yields consistent improvements. The recent fine-tuned external critic, however, does not generalize as reliably to the newly-introduced bias types (degradation observed in 5 out of 9 cases). Our findings demonstrate that prompting a judge to reason about bias, either preemptively or retrospectively, is a robust and generalizable approach for enhancing evaluation integrity.

0.23

Beyond Marginal Influence: Interaction-Aware Memory Selection for LLM Agents

cs.LG cs.AI ·2026-10-16 ·BHVSP Subrahmanyam, Neeraj Choudhary
benchmark
Abstract

LLM-agent memory selection often uses retrieval rankings based on record relevance or individual influence. Such rankings cannot identify sufficient subsets under redundant or jointly necessary evidence. We study per-decision memory sufficiency: selecting a small subset that preserves full-memory behavior. To separate search error from model error, we introduce JointCoreBench, a benchmark with nine interaction families, six workflow domains, natural-language decisions, and exact sufficient-set oracles, and JointCore, a relevance cost-ordered group-deletion baseline that tests blocks before individual records. On 6,912 exact conditions, cost-ordered JointCore recovers a minimum-cost sufficient set in 100.0% of cases, reduces subset evaluations from 30.00 to 11.13 (62.9%), and achieves 87.8% active-memory savings. Across four 12B-70B model families and 432 tasks per model, its unweighted mean is 10.28 versus 16.00 model queries and 50.6% versus 43.1% minimum-cost recovery relative to cost aware singleton deletion. Both methods preserve full-memory behavior under the deterministic primary interface. Tests with physical deletion, UNKNOWN masks, PENDING sentinels, counterfactual values, and alternative labels identify interface dependent behavior. JointCore is an efficient benchmark baseline; the study does not establish deployment-ready memory compression.

0.23

Reusing Calibration Gains across Model Representations

cs.LG cs.DB ·2026-10-16 ·Xu Cao, Wenxuan Xiao
Abstract

Scale calibration learns an update for one model representation. We study how much of that update can be reused by another representation of the same pretrained model. Our transfer procedure extracts positive group-wise gains and installs them in the recipient's weights or native quantization scales without recipient-side optimization. On a quantized recipient, it preserves integer codes, zero points and storage layout. For a Qwen3-1.7B floating-point/GPTQ INT3 pair, bidirectional transfer retains 76.9-90.5% of the text-loss improvement from recipient-specific calibration on fresh WikiText and C4 cohorts excluded from project development. Shuffling the same gain values within each projection worsens both recipients, showing that their learned group assignment matters after transfer. A Llama-3.2-3B control keeps the initial native tensors at all four calibrated projections identical and changes nine projections elsewhere. Cross-installation retains 98.6-99.3% of the text-calibration benefit, alongside a small, resolved preference for each recipient's own gain. These results quantify both the adaptation carried by a reusable calibration update and the remaining benefit of recipient-specific training.

0.23

GLOW: General Linear Models Optimized with Ward's Method

cs.LG ·2026-10-16 ·Matt Higger
unusual
Abstract

Neuroimaging seeks brain regions whose imaging features correlate with variables of interest (e.g., diagnosis, age), usually via the General Linear Model (GLM). Region-Based Analysis dilutes or splits any effect that does not fill exactly one atlas region. Voxel-Based Analysis, with Cluster Extent Thresholding and Threshold-Free Cluster Enhancement built on it, fits a GLM per voxel and groups adjacent significant voxels. But neighboring voxels can share a statistic without sharing a GLM, which a per-voxel statistic cannot detect. GLOW (General Linear Models Optimized with Ward's method) instead builds a hierarchy of nested regions by greedily minimizing GLM error (provably a special case of Ward's clustering). Maximum-statistic (Westfall-Young) permutation controls family-wise error over all regions. Significance spreads to an effect's nested fragments and the regions containing it; a final pruning step seeks whole, pure effect extents. We plant synthetic effects of known extent and varying strength in Human Connectome Project (HCP) diffusion images and in white noise. Pruning delivers wholeness: GLOW returns a spatially consistent effect as one region on 57% of HCP trials, the best baseline on 35%; baselines recover comparable volume, but in many pure pieces. The cost is positive predictive value (PPV): GLOW is more sensitive (by up to 0.29), but its regions overrun the effect, mostly because no hierarchy region isolates it, not because of pruning. On HCP, GLOW's mean Dice gains 0.059 over the best baseline at weaker effects and gives back 0.005 at stronger ones. On white noise, GLOW trails by up to 0.61 Dice: its hierarchy holds the effect, but its permutation test does not certify it. Runtime grows roughly linearly with voxel count, about 4.5 times the voxel-wise methods' on the full HCP volume. The PPV loss may trace to the assumption of spatially independent residuals.

0.23

Towards a GraphNARX model for structural dynamics

cs.LG ·2026-10-16 ·George Tsialiamanis, Keith Worden
unusual
Abstract

The current paper introduces the GraphNARX model, a graph-based nonlinear autoregressive exogenous model for structural dynamics. The model is built to account for the geometry of structures and to model dynamic behaviour as the interaction between elements of the structure. Similar to other types of NARX models, GraphNARX is used for system identification. Nodal attributes of the graph encode past displacements and nodal forces, and the output of the model is the next-timestep displacement. The graph formulation of the model allows it to generalise to structures of different dimensions and degrees of freedom. Moreover, because graph-learning is performed locally between the nodes, the GraphNARX model can also learn to infer displacements of nodes for which data are not available, i.e.\ to perform virtual sensing. The proposed model's performance is evaluated in a system-identification framework and a virtual-sensing framework, and achieves good accuracy and exhibits generalisation capabilities to structures of different sizes.

0.23

Sampling the XY model with an attention-based neural network

cs.LG cond-mat.dis-nncond-mat.stat-mechhep-lat ·2026-10-16 ·Piotr Białas, Christian D. Morales, Tomasz Stebel, Dawid Zapolski
Abstract · 19 pages, 6 figures

We present a transformer-based variational autoregressive network for sampling lattice systems with continuous degrees of freedom and apply it to the two-dimensional XY model near the Berezinskii-Kosterlitz-Thouless transition. The architecture is combined with physics-informed ingredients: locality-informed attention and approximate probabilities. The resulting sampler significantly outperforms existing neural samplers. We train the model on lattices up to $96\times96$, verify with a mode-dropping estimator that no parts of the configuration space are missed, and reproduce the energy, magnetization and helicity modulus obtained with the traditional Markov Chain Monte Carlo algorithm.

0.22

StoBasket: Joint Upsell and Cross-Sell Optimization for Contribution Margin

cs.AI ·2026-10-16 ·Xin Heng
unusual
Abstract

Retail recommenders rank relevance, but retailers earn contribution margin: an upsell earns the margin difference over the item it replaces, whereas a cross-sell earns its full margin. We introduce joint stochastic basket editing, a formulation in which one policy selects replacement and addition offers for an existing basket, values them by expected incremental contribution margin, and couples them through displayed-offer, accepted-offer, and joint-acceptance response channels. To our knowledge, it is the first formulation in which upsell and cross-sell responses are coupled; prior separable treatments are its zero-coupling special case. In real-data-calibrated simulations on five public basket datasets, StoBasket captures essentially all attainable offer margin: within 0.5% in every default-price setting and within 0.7% when the true response lies outside its model family. Correct economic accounting is the broadest gain: margin-weighted popularity leaves up to 11% of attainable margin unrealized, and a published next-basket recommender leaves 11-52%. Joint selection adds 10.2% of attainable margin on Instacart, where premium tiers and add-ons are strongly linked. When offers pay only together, the best heuristics lose 64% while exact search stays within 0.5%. For a structured subclass, we characterize when staged policies fail and give an exact optimizer that solves every tested real basket in 0.13 ms median and at most 111 ms. Profit saturates long before prediction accuracy: a correctly structured predictor trained on 5,000 randomized sessions comes within 0.06% of perfect prediction, while a signal-free one loses 38%. The same estimates yield a decision-aware market-basket analysis whose cost does not grow with association order.

0.22

Inducing Fourier Geometry: A Smoothness Prior Imposes Approximate Translation Equivariance in Multimodal Stellar Embeddings

cs.LG ·2026-10-16 ·Ilay Kamai, Alex M. Bronstein, Hagai B. Perets
unusual
Abstract

Learned representations often contain geometric structure induced by symmetries in the underlying data. We ask whether a known continuous parameter space can instead be used to \emph{induce} such structure when the corresponding symmetry is only approximate. We introduce a graph-Laplacian smoothness prior that encourages representations to vary smoothly over a physical parameter space without specifying a representation geometry; in the continuum limit it selects Laplacian eigenfunctions and therefore predicts the Fourier-like geometry associated with translation symmetry. On multimodal stellar representations spanning six astronomical surveys, the prior induces Toeplitz Gram structure and sinusoidal modes along stellar parameters, including in modalities whose native representations lack this geometry, while modality-specific residuals preserve structure outside the imposed coordinates. Of the exact-symmetry predictions, stationarity in a recoverable warped coordinate survives; diagnostics assuming exact quantization or asymptotic rates do not. The prior also improves downstream linear-probe performance and label efficiency, suggesting a general strategy: where a meaningful continuous parameter space exists, its local geometry can serve as a data-informed prior.

0.22

The Sign Code: The Hidden Binary Nature of Deep Networks

cs.LG ·2026-10-16 ·Niclas Göring, Shuofeng Zhang, Roi Holtzman, Yoonsoo Nam, Ard A. Louis
unusual
Abstract

Every neuron in a trained network outputs a number that can be separated into its sign (positive or negative) and its magnitude (absolute size). We call the set of all neuron signs the \textit{sign code}. This raises a key question: does classification depend more on the sign code or on the magnitudes? For overparameterized vision classifiers with sign-preserving activations, the answer is overwhelmingly the sign code. A $k$-nearest neighbor classifier operating only on this binary sign code achieves accuracy comparable to the full model across different architectures, activation functions, and datasets. Moreover, replacing all magnitudes in every layer with random noise while keeping the original signs barely changes residual network performance. In contrast, flipping just 20\% of the signs reduces accuracy to chance. Models generalize well when same-class inputs share similar sign codes. When this overlap is disrupted by a regularizer, training accuracy remains high, but test performance collapses. A margin decomposition helps explain these results: the sign-only contribution to each logit grows as $\sqrt{d}$ in network width, dominating the magnitude residual. The natural unit of analysis for these networks may not be through their continuous weight space but instead through their discrete sign codes.

0.22

Contracting with an Agent under Model Mismatch

cs.LG cs.GT ·2026-10-16 ·Elif Yılmaz, Jakub Tłuczek, Victor Villin, Christos Dimitrakakis
Abstract

In the classical principal-agent setting, the principal delegates some task to the agent, who receives a payment conditioned on the outcome of the task, which in turn benefits the principal. To align their objectives, the principal designs a contract that incentivizes the agent to take actions beneficial to the principal. In practice, however, the agent may act under misspecified beliefs about the underlying environment, leading to systematically suboptimal decisions. In this work, we study this problem through a principal-agent Markov Decision Process (PAMDP), in which the principal repeatedly interacts with an agent whose transition model may differ from the true environment dynamics. The principal has access to the true transition model and designs contracts that account for the agent's belief and incentives. We characterize contract design under model mismatch and develop a framework that enables the principal to induce desirable long-term benefits even though the agent has a potentially inaccurate understanding of the environment.

0.22

Uncertainty-aware AI-enabled framework for predicting hydrogen dispersion in confined spaces to enhance operator situational awareness

cs.LG ·2026-10-16 ·Niclas Flehmig, Mary Ann Lundteigen, Shen Yin
Abstract · Submitted to Journal of Loss Prevention in the Process Industries

We present a two-stage AI-enabled framework for enhancing operator situational awareness during compressed hydrogen bunkering in maritime confined spaces. In the event of a leak, the framework first estimates the unknown leakage rate from early sensor readings and then forecasts how the hydrogen cloud will disperse over the next time steps, providing operators with a predicted concentration map together with confidence bounds. The framework is based on two independent AI models building one AI system. The AI system is designed as a decision-support tool. It augments sensor readings, and thus enhances operator situational awareness and supports decision-making. This enables preventive safety barriers to be deployed before hazardous concentrations are reached, while an independent emergency shutdown system remains the ultimate safety layer. The system design incorporates safeguards to ensure the AI system operates within its competence envelope. Trained and tested on an open dataset of confined-space hydrogen releases, the framework achieves a mean absolute error of $0.052$~g/s in leakage-rate estimation and $\approx0.018$ in hydrogen-concentration prediction, approaching the sensor precision bound of $\pm0.01$. The hydrogen-concentration prediction provides well-calibrated uncertainty estimates that allow operators to judge the reliability of each forecast. These results suggest that the framework may provide early information about hazardous concentration build-up, potentially supporting preventive actions by operators.

0.22

A Training-Free Socket for DNA in Frozen Language Models

cs.AI q-bio.GN ·2026-10-16 ·Arash Fatehi
benchmark
Abstract

Connecting a DNA foundation model to a large language model (LLM) usually requires training a connector, and often the LLM too. We show that a closed-form map is enough for gene-level questions. Our method, the socket, fits one ridge regression (the plug) from frozen Evo 2 features to the host LLM's own representation of the gene's symbol, name, and description at an early layer. The plug's output replaces the hidden state at placeholder positions, and no gradient passes through the host. We test it on GeneHop, a five-way multiple-choice benchmark split by homology. In a single run on the held-out test set, under a frozen analysis plan, real DNA beats shuffled DNA on all four hosts: by +22.2 points [19.8, 25.0] on Qwen3.5-4B, +26.1 [24.3, 27.9] on Qwen3.6-27B, +16.7 [14.2, 19.7] on Mistral-7B, and +18.4 [16.2, 21.0] on OLMo-2-7B. All four survive Holm correction, every null control stays within two points of the sham level, and the development results agree within 2.3 points. A layer rule that uses no task items was selected on three hosts; it predicted the working layer of the fourth before any DNA-plug scoring on it. A text-only linear map lets one plug feed any other host, at a cost of at most 0.9 points on test. The socket conveys gene identity: nearest-neighbour retrieval recovers 72-84% of the effect, but the plug beats it on every host by 3.0-5.2 points and keeps most of its effect without training homologs.

0.22

From Flow-Box Invariants to Global Conservation Laws: Machine Learning-Assisted Detection of Singularities and Near-Integrability

cs.LG nlin.CD ·2026-10-16 ·Jimmie Adriazola, Ioannis G. Kevrekidis, Panayotis G. Kevrekidis, Wei Zhu
theory
Abstract

Conservation laws reveal invariant structure in dynamical systems and constrain their evolution to lower-dimensional level sets. Locally, however, per the flow-box theorem, every $d$-dimensional system generically admits $d-1$ functionally independent conservation laws near a regular point even though, generically, only a few such laws persist globally. Data-driven methods typically learn conservation laws on a prescribed region of phase space from which training samples are drawn, leaving their behavior beyond that region unresolved. We formulate this local-to-global continuation problem and develop an {expanding-domain neural deflation} framework that tracks the number of independent conservation laws over nested domains. The method recovers the expected $d-1$ local invariants and detects their successive breakdown as the domain expands, relaxing to the expected global conservation law limit. It distinguishes the local invariants (associated with Frobenius integrability) from persistent global structure and shows that the domain scale over which additional laws survive reflects proximity to Liouville integrability. Pointwise losses can localize isolated obstructions, whereas in more complicated dynamics the breakdown of our loss functions is gradual and spatially diffuse. The framework thus provides a data-driven approach for studying the persistence and breakdown of (so-far-learned) conservation laws across phase space.

0.22

How Many, but How Sure? Class Competition as a Confidence Proxy in Multi-Class Density-Based Counting

cs.LG ·2026-10-16 ·Villanelle O'Reilly, Georgios Leontidis, Marc Hanheide
unusual
Abstract · 2+1 pages, 1 figure; a 2027 Northern Lights Deep Learning Conference abstract

Multi-class density estimators can count objects in crowded scenes, but their class-wise outputs offer no indication of how reliably density mass has been allocated. We introduce a training-free confidence proxy based on local competition between the strongest class channels. The resulting prominence measure requires only the predicted density maps, with no retraining or access to model internals. We test whether residual competition is diagnostic of cross-class leakage by attenuating contested density at inference time. On DSACA predictions for the Hicks et al. dataset, this reduces overall MAE from 0.920 to 0.846, suggesting that class competition contains useful information about count reliability.

0.22

Joint, Decoupled, or Agent: When Each Multimodal Alignment Paradigm Wins

cs.AI cs.CLcs.CVcs.MM ·2026-10-16 ·Yi Fang, Gaurav Srivastava, Sirui Ding, Xuan Wang
Abstract

Jointly pretrained multimodal encoders are a common way to combine modalities, while alternatives integrate their information during downstream training or inference. When does each approach win? We compare three paradigms, instantiated with public checkpoints, across 16 paired datasets in five domains and multiple training-label budgets. JOINT trains a fusion head on frozen, jointly pretrained encoders. DECOUPLED trains a fusion head on frozen, independently pretrained encoders. AGENT gives a frozen LLM task instructions, labelled examples, available sample text, and supervised unimodal readouts. No paradigm dominates the complete 144 dataset-budget combinations: DECOUPLED wins 54, JOINT 44, and AGENT 46 by mean utility. ScienceQA, MELD, and MetMeme shift from AGENT to fusion as supervision increases. Between the fusion paradigms, JOINT leads on four of five QA datasets across all evaluated budgets, while higher unimodal 5-NN accuracy is associated with a relative shift toward DECOUPLED across eight classification datasets (Spearman $ρ=-0.67$). We test whether representation and task diagnostics predict these preferences on held-out datasets before fitting their candidate systems. A two-step cascade combines these diagnostics with task and budget metadata to select AGENT or fusion, then JOINT or DECOUPLED. Under leave-one-group-out evaluation, it selects the winner in 71.4% of cells versus 54.2% for metadata-only routing, with mean relative regret 0.028 versus 0.057. Using geometry, supervised representation diagnostics, and text-label matching together yields higher accuracy and lower regret than using any one of these sources with metadata. These results connect paradigm preferences to supervision budget, task format, and unimodal 5-NN accuracy. Target-task diagnostics thus support selecting among the three paradigms before candidate fitting.

0.22

AI-Driven Reproducibility and Analytical Robustness in Scientific Discovery

cs.AI cs.CL ·2026-10-16 ·Shicheng Liu, Krista Opsahl-Ong, Andrew Drozdov, Julian Killingback, Arnav Singhvi +4 more
benchmark
Abstract

Reproducible and analytically robust research findings are essential to scientific progress, but evaluating these properties typically requires costly manual effort. We investigate whether LLM agents can help automate reproduction and robustness analysis, using a medical case study grounded in the Surveillance, Epidemiology, and End Results (SEER) database, a widely used U.S. cancer registry underlying tens of thousands of published studies. First, we introduce SEER-Repro, a benchmark of 20 published SEER-based papers together with the data and resources required to reproduce their findings. Second, we create an AI agent that successfully reproduces 77.4% of 2,190 atomic claims, while providing executable baselines for further analysis. Third, we develop an agentic analysis pipeline to test the robustness of the medical findings by identifying analytical choices underlying each finding, proposing defensible alternatives, and rerunning the analysis to determine whether those choices affect the conclusion. Across published papers in SEER-Repro, our pipeline identifies potentially consequential analytical choices in 9.5% of fully reproduced findings and flags them for expert review. These results show how LLM agents can support scientific review by reproducing findings, systematically testing their robustness, and directing expert attention toward potentially consequential analytical choices.

0.22

Bootstrap Stochastic Interpolant Tree for Inference-time Search Under Sparse Reward

cs.LG cs.AI ·2026-10-16 ·Binglin Ji, Anindya Sarkar, Yevgeniy Vorobeychik
Abstract · 26 pages, 12 figures, Preprint

Tree-based samplers offer a compelling alternative for scalable inference-time search, bypassing the per-step reward and gradient dependencies that introduce systematic bias into SMC-style frameworks. Despite this advantage, contemporary tree samplers fail to achieve broad exploration. Their reliance on learned, DDPM-like transitions inherently couples neighbor states in a trajectory, thereby restricting child-parent feature divergence in the tree sampler. Consequently, these samplers suffer from restricted coverage of the target marginal distribution, severely hindering the exploration critical for efficient search under sparse reward. To overcome these limitations, we introduce a simple yet effective Bootstrap Stochastic Interpolant Tree (BSIT) sampler that decouples child node generation from explicit parent dependencies. By expanding the degrees of freedom during child node sampling, BSIT encourages diverse trajectory rollouts and unlocks broader path-space diversity-the exact catalyst needed for efficient exploration. Crucially, we show that BSIT's improved exploration accelerates convergence to the target distribution while achieving broader distributional coverage than samplers that rely on DDPM-like transitions for child-node generation, all without incurring additional function evaluations or reward feedback, thereby providing a highly efficient basis for rare target search. Extensive experiments across diverse search settings, including rare-species discovery and target-aligned molecule generation tasks, demonstrate BSIT's superior search efficiency over the baselines.

0.22

HiPredict: Predicting the missing HiRISE RED4 spectral channel with machine learning

cs.LG ·2026-10-16 ·Ardan Suphi, Ben Moseley, Gareth S Collins, Valentin T Bickel, Vidhya G Rangarajan
Abstract

The High Resolution Imaging Science Experiment (HiRISE) onboard the Mars Reconnaissance Orbiter (MRO) acquires imagery of Mars at 25--30 cm/pixel, supporting studies of martian geology, climate history, and landing-site characterization. Its 14-detector focal plane assembly includes several overlapping detectors that enable the acquisition of three-channel colour-infrared images, which are important for identifying compositional variations and studying active surface processes. In 2023, a detector in the center of the focal plane assembly (RED4) began failing intermittently, creating a gap in acquired imagery and reducing the instrument's colour-imaging capability by half. The HiRISE processing pipeline provides a synthetic red (SYN4) channel obtained by linearly scaling the co-located infrared (IR10) channel. However, this approach cannot capture observed non-linear relationships between the colour channels, limiting its accuracy and applicability. Here, we present HiPredict, a machine-learning framework that reconstructs RED4 from co-located IR10 and blue-green (BG12) imagery, neighbouring red-channel observations, and acquisition metadata. The model employs a U-Net architecture to learn non-linear spatiotemporal-spectral relationships between these inputs and is trained on 81,717 training samples, representing 13,625 observations. Evaluated on 23,348 test samples, representing 3,892 pre-failure observations, HiPredict reconstructs RED4 with a Normalised Mean Absolute Error of 1.1%, substantially outperforming SYN4 (3.1%). The reconstructed products improve colour image recovery and more closely represent features of interest such as icy craters, stratigraphy exposed along ridge crests and chloride deposits. We release HiPredict as an open software tool. This work demonstrates the potential of machine learning to mitigate data loss as planetary missions age and naturally degrade.

0.22

Thought Blocks: Non-Autoregressive Latent Communication for Multi-Agent Systems via Trajectory-to-State Compilation

cs.MA ·2026-10-16 ·Zhenhan Guan, Xiaolong Liu, Hengwei Ye, Zheng Tian
Abstract

LLM-based multi-agent systems typically communicate through text, incurring repeated generation and re-encoding costs. Although latent communication removes explicit verbalization, constructing latent messages may still require autoregressive latent trajectories. We therefore distinguish message representation from message construction and ask whether agents can directly construct useful continuation states without executing the trajectories that normally produce them. We formulate this problem as trajectory-to-state compilation and introduce Thought Blocks, fixed-size blocks of latent key--value (KV) continuation memory constructed in one shot. Thought Blocks achieve a 2.5 times average latency speedup over autoregressive latent communication. Compared with text-based multi-agent communication, Thought Blocks reduce text tokens by 45.7% and autoregressive steps by 84.8%, achieving an 8.6 times average latency speedup. These results demonstrate that trajectory-to-state compilation enables efficient inter-agent latent communication without sequential latent rollouts at inference.

0.22

Graph-Regularized Synthetic Control for Stable Counterfactual Estimation under Donor Redundancy

cs.LG cs.AIecon.EMstat.ME+1 ·2026-10-16 ·Yanlin Zhang, Sungyong Chung, Alireza Talebpour
benchmark
Abstract

Synthetic control methods estimate counterfactual outcomes by weighting untreated donor units to match a treated unit before treatment. In recent applications, however, the donor pool can be large, highly correlated, and short in pre-treatment history, yielding many near-equivalent solutions with unstable and difficult-to-interpret donor weights. We propose Graph-Regularized Synthetic Control (Graph-SC), a convex estimator that augments the classical pre-treatment matching objective with a graph Laplacian penalty. The graph encodes similarity among donor trajectories, encouraging coherent weights over structurally related donors while preserving the transparency of synthetic controls. We provide theoretical results characterizing the bias-variance tradeoff induced by graph regularization and establish efficient convergence guarantees with an accelerated proximal gradient method. Across a controlled simulation under donor redundancy, the California Proposition 99 benchmark, and observational counterfactual analyses on the Waymo Open Motion Dataset, Graph-SC improves weight stability, reduces sensitivity to redundant donor pools, and maintains competitive pre- and post-treatment accuracy relative to classical and regularized synthetic control baselines. Our results suggest that graph regularization is most valuable not as a universal accuracy booster, but as a principled mechanism for selecting stable and interpretable counterfactuals from near-equivalent synthetic controls with comparable pre-treatment fit.

0.22

No-Regret Language Generation

cs.LG stat.ML ·2026-10-16 ·Jiaxun Li, Vinod Raman, Ambuj Tewari
Abstract

We introduce no-regret generation, which requires a generator to make a sublinear number of mistakes uniformly over all demonstration sequences and compatible target languages. We also formulate an agnostic variant that allows arbitrary demonstrations that may not be compatible with any languages. For both settings, we introduce tree-based mistake-forcing dimensions that exactly characterize finite-horizon minimax complexity and establish necessary and sufficient conditions for no-regret generation. Our characterizations reveal a rich spectrum of generation rates: up to constant rescaling, every admissible nondecreasing profile is attainable in both the realizable and agnostic settings. Finally, we connect generation complexity to online prediction: for language classes of finite Littlestone dimension \(d\), we obtain realizable mistake bounds of \(O(d\log(1+T/d))\) and agnostic regret bounds of \(O(\sqrt{dT\log(1+T/d)})\). The realizable bound is optimal up to universal constants.

0.21

Torch-DPhys: A Torch-Native, MOOSE-Compatible Framework for Differentiable Physics Simulation and Hybrid Machine Learning Integration

cs.LG ·2026-10-16 ·Akzhol Almukhametov, Yang Liu
Abstract

We present Torch-DPhys, a Torch-native framework implemented in PyTorch for differentiable physics simulation and hybrid physics-machine learning. It implements selected thermal-hydraulic modeling concepts and discretizations from MOOSE's Thermal Hydraulics Module and the System Analysis Module (SAM), with a preliminary extension to continuum mechanics. Native tensor residuals support CPU and GPU execution, compilation of selected numerical kernels, and sparse direct linear solves. Sensitivities are obtained by differentiating executed numerical operations or through discrete adjoints of converged residual equations. Verification combines analytical solutions, reference-code comparisons, and derivative checks. Demonstrations cover plane-strain elasticity, one- and two-loop thermal-hydraulic transients, and an axisymmetric molten-salt fast reactor. For the reactor, comparison with SAM-Lite gives a maximum normalized field discrepancy of 1.13 x 10^-5 over 200 adaptive time steps, demonstrating agreement between the tested discrete implementations. Selected transient sensitivities are assessed using tangent-linear calculations and finite differences on fixed time schedules. Two applications illustrate the framework's connection between simulation and optimization. A residual-adjoint parameter-identification calculation recovers nominal pump head and heating power with relative errors below 0.06% from noiseless synthetic MOOSE observations. A separate three-loop plant demonstration uses solver-in-the-loop training of a predictive controller by differentiating trajectory-level objectives through the thermal-hydraulic simulation. These examples illustrate a foundation for gradient-based inference, control, and coupled physics-machine learning models in nuclear engineering.

0.21

Local proxies for memory-efficient training

cs.LG cs.AI ·2026-10-16 ·Wu S. Zihan, Brian A. Yuan, Guillaume Bellec
Abstract

Local learning algorithms can reshape resource requirements of artificial intelligence. The memory bottleneck of standard end-to-end backpropagation lies in the need to store all network activations until the loss gradient has been propagated back through. Local learning avoids this, but typically at a cost in accuracy. We study how to achieve accurate and memory efficient local learning in transformers. For each transformer block, we train a local proxy network to predict the global network output via distillation. Gradients can be independently propagated through this local proxy while estimating well the backpropagation gradients. In theory, this approach is exact for deep linear networks; in non-linear residual models, the gradient error can be bounded when using the appropriate residual proxy. We verify this numerically and demonstrate that optimizing the distillation loss improves gradient estimation. Evaluating the performance of proxy propagation on CIFAR-100 in vision transformers, we find that it achieves a better accuracy/memory trade-off compared to other local algorithms. Finally, we show that the method is further relevant for fine-tuning large language models on a small memory budget. Our method trains faster and achieves higher performance than memory-efficient zeroth-order learning algorithms with comparable memory cost.

0.21

LogicVal:A Logic-Formalized Benchmark for Constraint Satisfaction in Text-to-Image Generation

cs.AI cs.CLcs.LO ·2026-10-16 ·Savitha Sam Abraham, Marjan Alirezaie, Hossein Rahnama
benchmark
Abstract · Preprint. Under review. Savitha Sam Abraham and Marjan Alirezaie contributed equally

Text-to-Image (T2I) generative models have made significant progress in producing high-fidelity images from natural-language prompts. Alongside this progress, numerous benchmarks have been developed to evaluate capabilities such as attribute binding, numeracy, spatial reasoning, and commonsense reasoning. In this work, we propose a new perspective on T2I evaluation by casting it as a constraint-satisfaction problem. We introduce a logical framework that formally specifies the benchmark domain, including objects, attributes, and relations, alongside the constraints that generated images must satisfy. We evaluated four state-of-the-art T2I models across 9 classes of constraints ranging from a single-object-attribute to multi-object relationships constraints involving logical operators ($\forall, \exists, count, \rightarrow$). Our results reveal a substantial gap between single-object and complex conditional multi-object relationship constraint classes, with the best-performing model achieving 98% accuracy on the former and just 53% on the latter. We assess compositional generalisation, finding that the best-performing model`s accuracy drops from 84% on single-constraint prompts to 20% when satisfying multiple constraints simultaneously. Identifying unsatisfiable constraints is also challenging, with the best model correctly recognizing only 23% of such prompts. Building on this framework, we introduce \textbf{LogicVal}, an evaluation methodology that combines neural visual detection with logical reasoning to assess constraint satisfaction in generated images. On CLEVR, LogicVal achieves ROC-AUC scores ranging from 0.88 to 0.98, indicating strong alignment with human judgments. These findings underscore the value of logic-based evaluation for characterizing T2I model capabilities and limitations.

0.21

Optimist Deep Research: An Iterative, State-Centric Agent for Autonomous Literature Research

cs.AI ·2026-10-16 ·Gihan Panapitiya
code/data
Abstract

Most retrieval-augmented systems treat literature search as a single retrieval step followed by summarization. Real literature research works differently: the investigator's understanding of the question changes as they read, new subquestions appear, and evidence is weighed, not just collected. We present Optimist Deep Research, an open-source agent that runs literature research as an iterative evidence-acquisition process over an explicit, mutable research state. A research question is broken into a tree of subquestions, each served by search intents that are translated into source-specific queries over arXiv, OpenAlex, Crossref, PubMed, Semantic Scholar and the web. Results are merged into a canonical, provenance-preserving document registry through a conservative four-tier deduplication hierarchy. The registry is expanded by citation snowballing until a saturation criterion is met, and then screened, ranked and allocated to subquestions round-robin. Documents are read in parallel into a structured evidence store of claims with support, location, type and strength. A critic reviews the state along eight dimensions and returns research actions rather than a binary verdict: it can add, split, rephrase, deprioritize or close subquestions, so the research plan is an output of the investigation as well as an input. The final report is written section by section from the evidence store, and every citation resolves to a retrieved document. We describe the design, the deterministic machinery that keeps a language-model pipeline auditable, and a case study that read 28 of 237 retrieved documents in about three and a half minutes. We also report weaknesses the case study exposed.

0.21

SteerSearch: Interactive Search Reasoning with Mid-Response Steering

cs.AI ·2026-10-16 ·Udit Thakur, Baohao Liao, Matteo Nulli, Stefan Vasilev, Seyyed Hadi Hashemi +2 more
benchmark
Abstract · 48 pages

Search-augmented language models increasingly serve as long-horizon research agents, yet most open training methods assume a fixed user goal throughout search. We study mid-response steering, where users redirect a model during ongoing reasoning or retrieval. We identify answer-target anchoring, a failure mode in which fixed-query search agents retrieve evidence for the updated request but still answer an earlier question. We introduce SteerSearch, an RL framework that trains on interactive search trajectories with injected interventions to enable context-preserving target reassignment, and SteerSearch-Bench, a benchmark adapted from multi-hop QA datasets. We also investigate Contrastive Context Reward (CCR), an auxiliary signal for aligning post-intervention reasoning with the updated goal. Without CCR, SteerSearch improves average steering accuracy over Search-R1 from 11.1% to 31.3% on Qwen2.5-3B and from 36.6% to 44.6% on Qwen3-4B, while preserving standard single-query performance. Gains also extend to larger Qwen2.5 models and natural multi-turn conversations.

0.21

Logic-Guided Graph Denoising Diffusion Models

cs.LG ·2026-10-16 ·Samuel Cognolato, Luciano Serafini, Alessandro Sperduti, Andrea Cavallaro
Abstract · Manuscript under review at AAAI2027

Graph diffusion models are competitive approaches for generating complex graph-structured data. However, when such distributions have strict structural constraints, diffusion models may fail to capture them exactly, resulting in the generation of invalid graphs. In this paper, we propose a method for grounding graph diffusion models in domain-specific knowledge via First Order Logic (FOL), leveraging its expressive and interpretable formalism to impose structural constraints on the generative process. We incorporate domain knowledge using classifier guidance, a technique that conditions the diffusion model's distribution on external variables to exert control over the sampled data. During the generative process, classifier guidance steers the denoising trajectory to optimize an objective function, which we define as a fuzzy and differentiable relaxation of the logical rules. Our methodology requires only the specification of constraints in FOL, whose satisfaction is encouraged at inference time to guide a pre-trained generative model. We evaluate our approach by comparing the unconditional generation of knowledge graphs with our logic-guided framework, which incorporates domain ontologies as structural constraints during inference. This approach significantly increases the proportion of generated graphs that satisfy the ontological constraints.

0.21

Visual Causal-chain Bookmarking: Retrieval-Weighted Credit Assignment for Reinforcement Learning from Delayed Reward

cs.LG ·2026-10-16 ·Prathap Selvakumar
Abstract · 9 pages, 2 figures, 5 tables. Master's dissertation research, University of Manchester

RL agents trained on a single trajectory-level return face a credit-assignment problem: when a reward arrives many steps after the decision that caused it, every action gets the same signal regardless of causal responsibility. This paper studies Visual Causal-chain Bookmarking (VCBM), which concentrates credit on the decision points that determine an episode's return, via two instantiations. The first targets a fully observed, low-dimensional MDP (a delayed-reward watch-repair task), where VCBM statistically discovers which state components the agent causally controls and estimates per-decision-point action values directly, evaluated against ground truth and RUDDER, an LSTM-based baseline. VCBM converges to 95.07% correct decisions in a mean of 1,532.3 episodes across 20 seeds, versus RUDDER's slower, more variable convergence (mean 3,327.1 episodes; p<0.001); VCBM's discovered causal structure exactly matches ground truth in all seeds. The second targets a partially observed, language-based setting: an LLM agent trained with Leave-One-Out PPO (LOOP) in AppWorld, where causal discovery is intractable and a structural heuristic (state-mutating tool calls) substitutes for it, with credit distributed via retrieval over episodic memory and blended with LOOP's advantage. Over 200 iterations on identical Qwen2.5-7B-Instruct/LoRA/FSDP2 configurations, VCBM-blended LOOP improves task success over plain LOOP (0.354 vs. 0.458), reduces final-step max importance weight, and improves token efficiency, with a small KL exception favouring LOOP. The watch-repair result supports the claim that weighting causally relevant steps is more sample-efficient than an undifferentiated baseline or a learned return-redistribution network; the AppWorld result, limited to one seed per arm by a 2-GPU budget, is a compute-limited demonstration at a harder, natural-language scale, not a statistical confirmation.

0.20

LagNet: Lag-Structured Operators for Long-Term Time Series Forecasting

cs.LG ·2026-10-16 ·Zhenan Yu, Guangxin Jiang, Jiuyun Jiang, Jin Yang
Abstract

Long-term time series forecasting maps a finite history window to an extended future horizon. Many existing models focus primarily on the design of input representations and model structures, while paying less attention to the structural form of the history-to-future forecasting operator itself. We examine this operator directly and find that, across model families, sensitivity matrices exhibit recurring diagonal patterns whose diagonals correspond to relative temporal lags. Motivated by this observation, we develop a framework for direct multi-step forecasting based on lag-structured temporal operators. Its basic building block, LagLayer, constructs a Toeplitz matrix by tying history-to-future connections with the same relative lag and incorporates horizon-wise and history-wise scaling coefficients to relax exact sharing. This design enables LagLayer to serve as an effective replacement for temporal linear layers in existing forecasting models. Stacking LagLayer modules with activations and multiple feature maps yields LagNet. Experiments on long-term time series forecasting benchmarks show that our method is competitive with strong baselines, while sensitivity matrix diagnostics and limited-data experiments demonstrate that LagLayer provides an effective and sample-efficient alternative to a linear layer.

0.20

S-GAP: Stealthy Graph Attack Primitives for Graph Large Language Models

cs.AI ·2026-10-16 ·Wenda Li, Tongya Zheng, Shunyu Liu, Hanyang Yuan, Kaixuan Chen +6 more
Abstract · 8 pages, 4 figures, 3 tables

Graph Large Language Models (GraphLLMs) enable powerful semantic understanding and reasoning over Text-Attributed Graphs (TAGs), while their semantic reasoning may exhibit vulnerabilities to perturbations in node text and graph structure under adversarial attack evaluation. Existing attacks typically construct textual and structural perturbations in separate stages guided by distinct realization intents. However, this separation results in attack-intent fragmentation, causing semantic incoherence between the two perturbation types, thereby undermining attack stealth and making them easier for GraphLLMs to identify and disregard, ultimately weakening attack effectiveness. To avoid attack-intent fragmentation, we propose the Graph Attack Primitive (GAP), an atomic semantic representation that unifies textual and structural perturbations under a coherent attack intent. By carefully calibrating the semantic composition of a GAP, we can achieve strong attack effectiveness while maintaining sufficient attack stealth. Furthermore, we introduce S-GAP (Stealthy GAP), a multi-agent framework that iteratively refines the semantic content of the GAP. The framework unifies GAP planning and realization and continuously refines this process through surrogate evaluation and textual feedback. Extensive experiments across four datasets, three GraphLLMs, and two attack settings show that S-GAP improves ASR over the strongest baselines by 7.02% on average while producing more relationally plausible perturbations, exposing critical vulnerabilities in current GraphLLMs and motivating more robust defenses.

0.20

Blitz: Accelerating the Diffusion Modules of AlphaFold3 and Boltz-2

cs.LG ·2026-10-16 ·Vinay Saji Mathew, Soundar Kumara, Gretta Kellogg, William K. M. Lai
Abstract · 24 pages, 12 figures

Diffusion-based protein structure prediction models produce all-atom structures in about 200 denoising steps, and that cost has driven a number of few-step samplers. We present Blitz, a few-step distillation framework that uses a geometric noise schedule to distribute steps more evenly across noise scales, then learns those updates through multistep consistency training. Physical steering and target-free candidate selection help reduce the remaining geometric defects at inference. We apply Blitz to AlphaFold 3 and Boltz-2, reducing denoising from 200 steps to eight. Because published few-step results use different benchmarks and scoring procedures, we compare both distilled models with their teachers and other few-step methods under a common evaluation protocol covering structural accuracy, interface quality, ligand placement, and stereochemical validity. Blitz retains teacher-level structural accuracy, with AF3-Blitz outperforming all evaluated few-step baselines and exceeding its teacher on several accuracy metrics.

0.20

A Label Noise Robust Algorithm for Graph Data using Neighborhood Label Distribution

cs.LG cs.AImath.ST ·2026-10-16 ·Pintu Kumar, Nandyala Hemachandra
Abstract

Graph Convolutional Networks (GCNs) have demonstrated exceptional performance on graph-structured data. However, their success heavily relies on the availability of clean labels, an assumption that often fails in real-world scenarios. The learnability of GCNs is intrinsically linked to the neighborhood structure of nodes, making it critical to understand how this structure changes in the presence of label noise. In this work, we investigate how the neighborhood label distribution of a node changes under label noise. We mathematically derive that label noise smooths this distribution, which inherently increases its entropy and severely decreases interclass distinguishability. To address this challenge, we propose Leave-one-out Entropy Node Screening (LENS), an end-to-end framework that identifies noisy nodes by measuring the reduction in neighborhood entropy when a specific node is removed. LENS further incorporates a relabeling mechanism to recover and preserve valuable feature information instead of entirely discarding the identified noisy nodes. Crucially, the end-to-end computational overhead of LENS is remarkably low. Extensive evaluations demonstrate that our algorithm achieves state-of-the-art node classification accuracy while executing an order of magnitude faster than existing noise-robust approaches.

0.20

Latent State Utilization in Probabilistic Circuits

cs.LG ·2026-10-16 ·Zilei Shao, Anji Liu
theory
Abstract

Probabilistic circuits (PCs) are generative models that support efficient probabilistic inference. Adding latent states to a PC increases its representational capacity, but yields diminishing gains in likelihood. We use mutual information to examine how the dependence represented by a trained PC changes with latent size. Experiments with Hidden Chow-Liu Trees (HCLTs) on ImageNet32 show that learned dependence grows slowly and remains well below both the latent-state upper bound and estimates of dependence in the data. To understand how learning uses these states, we examine how strongly each sample favors particular latent states and how broadly states are used across the dataset. Tracking these measurements during HCLT training distinguishes rapid early posterior concentration from the more gradual stabilization of sample-dependent state preferences. Early measurements also guide initialization selection in HCLTs and hidden Markov models (HMMs), as well as tree selection in HCLTs. The resulting choices improve the ultimate performance across the tested latent sizes, providing a practical benefit from our study of latent-state utilization.

0.20

Compression Precedes, Content Decides: Emergent Misalignment in Reward Models

cs.AI ·2026-10-16 ·Dhruv Puri, Nikhil Prakash
Abstract

Emergent Misalignment - the out of domain generalization of narrow harmful fine-tuning first shown by Betley et al. [2025] has been widely studied in language models but not in reward models that supervise them. In this paper, we fine-tune two state-of-the-art reward models, Skywork-Reward-V2-Llama-3.1-8B and Skywork-Reward-V2-Qwen3-8B [Liu et al., 2025] and show that narrow fine-tuning reward models to prefer harm in one domain (medical or financial) generalizes to seven held out domains. This is not degeneracy, as the fine-tuned reward models lose only 4.8 and 5.3 points overall (RM-Llama) on the standard RewardBench 2 evaluation [Malik et al., 2025] while still outperforming the unmodified RM-Qwen model. We then fine-tune the underlying language models, Llama-3.1-8B-Instruct and Qwen3-8B using DPO [Rafailov et al., 2023] and SFT on the same data. We find that reward models are more vulnerable to preference inversions compared to their language model counterparts. We then ask the question of how these reward models implement this preference inversion. Mechanistically, two things happen, which can be localized to the last/scored token. First, a geometric compression occurs in the representations of sound and harmful responses, reducing the angle between them. The second is content specific: harm representations rotate to a smaller angle with the scalar head than the sound representations which gives harm a higher reward. This happens only in harmful preference fine-tunes and not in control.

0.20

LED: Improving Masked Diffusion Language Models by Lookahead Editing

cs.LG cs.CL ·2026-10-16 ·Yuyan Zhou, Zeyuan Yang, Yuxuan Song, Zhicheng Cai, Hongli Yu +3 more
Abstract · 10 pages, 5 figures

Diffusion large language models (dLLMs) have emerged as a promising alternative to autoregressive language models, enabling non-autoregressive text generation through parallel decoding. Existing dLLMs can be broadly categorized into two families: masked diffusion models (MDMs) and uniform diffusion models (UDMs) In this work, we theoretically show that MDMs are easier to optimize due to their mask-based transition schedule but have lower computational expressivity, whereas UDMs enable iterative state revision and greater expressivity but are substantially harder to train. This trade-off raises a natural question: can we obtain the expressivity of iterative revision without giving up the simple, easy-to-optimize transition schedule of MDMs? Motivated by this analysis, we propose Lookahead Editing (\ours{}), which augments masked diffusion decoding with a rewritable proposal state. At each decoding step, the model performs several lookahead updates to iteratively refine future predictions in this proposal state before committing tokens to the masked sequence. This design introduces a rewritable proposal that uses additional compute to improve expressivity while keeping the simple mask-based transition schedule of MDMs. We fine-tune Qwen2.5-7B and Qwen3-8B to obtain LED-\texttt{7B} and LED-\texttt{8B}, respectively. Across eight benchmarks, LED SFT yields average gains of $+11.4$ and $+7.7$ points over standard MDM SFT. Moreover, using SFT alone, LED-\texttt{8B} outperforms LLaDA2.1-16B Q-mode, which is trained with continued pre-training, supervised fine-tuning, and reinforcement learning.

0.20

A Topological Signature of Learning under Noisy Labels

cs.LG cs.AI ·2026-10-16 ·Zhongtian Sun, Fan Mo, Prayag Tiwari, Kelin Xia
Abstract

How does noisy supervision change the representations learned by a neural network? Training accuracy measures agreement with training labels but provides little information about representation geometry. We use persistent homology to track loop structure across distance thresholds in neural activations. Experiments span image classification, text classification, and modular arithmetic. On a fixed CIFAR-10/ResNet-18 training probe, equally sized clean-label and corrupted-label subsets show no detectable persistence difference at epoch 10, whereas corrupted-label subsets have higher persistence at epoch 50. The contrast survives uniform-scale controls, even though predictions match the assigned noisy labels for only 7-11% of corrupted probe examples at that checkpoint. Normalisation preserves the crossover between clean and noisy training trajectories on CIFAR-10/ResNet-18, whereas CIFAR-100/ResNet-18 shows no crossover under any of the five evaluated constructions. In a separate noise-rate calibration comparison across five pipelines, mean activation norm achieves a higher mean in-sample R2 than raw persistence (0.936 versus 0.908). These findings show that noisy supervision leaves a measurable geometric signature even when agreement with corrupted labels is low. Comparing persistence under scale controls with label-fitting measurements offers a way to study how representations respond to label corruption beyond what training accuracy reveals.

0.20

FLARE: Physics-Grounded Evaluation of Language-Model Agents for Feeder Restoration with Degraded Observations

cs.LG ·2026-10-16 ·BHVSP Subrahmanyam, Neeraj Choudhary
benchmark
Abstract

Restoring service after distribution-grid faults requires decisions from incomplete and conflicting information while respecting electrical constraints. Addressing this need, we introduce FLARE (Feeder-Level Agent Restoration Evaluation), a physics-grounded benchmark for post-isolation feeder restoration. FLARE con- tains 800 AC-feasible cases across five observation settings and allows agents to gather evidence, evaluate switches, and perform restoration actions under fixed operational limits. We evaluate five instruction-tuned models over 78,750 episodes, including topology, distributed-generation, and loading shifts. Mean normalized regret ranges from 0.082 to 0.148. Although every model has zero median regret, P95 regret ranges from 0.814 to 1.000, revealing an important failure tail. Shadow replay shows that 96.4% of 15,706 guard-rejected actions would have caused a physical violation. FLARE enables reproducible evaluation of language-model based restoration support while emphasizing the need for external safety checks.

0.20

Looped GNNs Beyond Fixed Depth: Graph Size Generalization and Parameter-Efficient Adaptation

cs.LG ·2026-10-16 ·Zhiyang Wang, Zhishang Luo, Yusu Wang
Abstract · 36 pages, 12 figures

Generalizing from small graphs to larger ones can require computation beyond the depth of a neural model used during training. We study target tasks that admit a common local iterative computation and whose effective dependency radius grows with graph size. The central idea is to learn a local update rule through GNN pretraining, then repeatedly apply the same learned processor beyond the training depth to extend information reach on larger graphs without retraining. Calling such parameter-shared architecture a \emph{looped GNN}, we categorize six schemes for looped GNNs to separate the effects of depth extension, fine-tuning, and freeze-and-steer adaptation. Our analysis establishes fixed-depth information limits and sufficient conditions under which accurate iterative reuse overcomes them. To improve beyond unadapted reuse, we introduce \emph{Loop-Freeze-Steer} (LFS), which freezes the pretrained processor and learns a small controller to steer its repeated application, using much fewer trainable parameters than full fine-tuning. Our finite-sample analysis gives sufficient conditions for adaptation to improve performance on larger graphs. Experiments on synthetic tasks and chip placement evaluate depth extension and adaptation, highlighting LFS's accuracy--trainable-parameter trade-off for size generalization.

0.20

Adaptive conformal prediction on the Grassmannian

cs.LG stat.ML ·2026-10-16 ·Dimitrios Loukrezis, Zachary J. Grey, Dimitris G. Giovanis
Abstract · 21 pages, 6 figures, 3 tables

We introduce a conformal prediction framework for regression with Grassmann-valued responses. The proposed method constructs prediction regions in the form of metric balls directly on the Grassmann manifold, using basis-invariant geometric distances as non-conformity measures and scaling the corresponding non-conformity scores through a learned prediction difficulty estimator. This yields locally adaptive, input-dependent prediction regions that respect the underlying manifold geometry, while retaining the finite-sample marginal coverage guarantee of split conformal prediction under exchangeability. We evaluate several Grassmannian distance metrics and compare their prediction-region efficiency on synthetic benchmarks and an airfoil inverse shape-design problem. The results highlight the advantages of geometry-aware non-conformity scores, while local adaptation improves empirical coverage uniformity across prediction difficulty levels and, in strongly heteroscedastic settings, can substantially reduce prediction-region size while maintaining the target coverage.

0.20

Reward Engine: Reward-Centric Self-Evolution for Open-Ended Tool Agents

cs.AI ·2026-10-16 ·Litong Feng, Ji Qi, Yiran Cheng, Shaowu Pan, Xinyu Zhang +1 more
benchmark
Abstract · 29 pages

Open-ended tool agents can complete the same objective through different tool paths, and their quality depends jointly on the interaction trace, environment state, and final artifacts, so such tasks rarely admit an executable ground-truth reward. The central challenge is therefore to construct a reward from existing experience that exposes the current policy's failures while remaining stationary during optimization. We introduce Reward Engine, which, like a self-motivated student, first works out what success requires and then optimizes toward it. From the task and initial trajectories, it builds a RewardPack of testcases for required correctness and rubrics for graded quality, each executed as a check that returns a judgment with evidence. The frozen RewardPack then guides the agent to revise and select its outputs, and can also serve as a reward for reinforcement learning. On WorkBuddy Bench Office and Claw-Eval, the constructed reward largely agrees with the benchmark graders, not reversing their ordering on 82.3\% and 80.5\% of candidate pairs. Used to revise executions, it raises three-run consistency ($\mathrm{Pass}^{3}$) over Baseline from 45.0\% to 54.0\% and from 78.4\% to 83.7\%.

0.19

PAC Learning with Query Feedback

cs.LG ·2026-10-16 ·Chase Hutton, Adam Melrod, Han Shao
theory
Abstract

We study realizable multiclass PAC learning with query feedback, where each training example provides a single binary response indicating whether its true label belongs to a chosen subset from a fixed query family. Our main contribution is the query dimension, a combinatorial parameter that captures both the structure of the hypothesis class and the ability of the available queries to distinguish labels. This parameter generalizes the bandit DS dimension, which it recovers under singleton queries. We establish a sample-complexity lower bound of $Ω(D/\varepsilon)$ and an upper bound of $\widetilde{O}(D^2/\varepsilon)$, suppressing factors polylogarithmic in the number of labels and the failure probability. When the available queries can split every candidate set into two parts of comparable size, we improve the upper bound to $\widetilde{O}(D/\varepsilon)$, yielding a characterization up to logarithmic factors.

0.19

Minimax-optimal Scaling Laws in Ridge Regression via Transfer Learning

cs.LG stat.ML ·2026-10-16 ·Yalda Shabanzadeh, Diyuan Wu, Marco Mondelli
hot direction
Abstract

Pretrained models are often reused across downstream tasks, motivating the question of whether this transfer learning procedure can provide a substantial performance improvement, namely, in terms of the scaling law. We address this question in the prototypical setting of high-dimensional regression, where (i) a pretrained \emph{source} model is trained via ridge regression, and (ii) a fine-tuned \emph{target} model is regularized via the $L_2$ distance from the source estimator. First, we establish a deterministic equivalent for the risk of this two-step procedure under both model and covariance shift. Next, by specializing the deterministic equivalent to a shared regression function and commuting power-law covariance spectra, we identify a regime where the resulting scaling law achieves the minimax-optimal exponent. Notably, in this regime, the optimal source regularization depends on source data only, indicating its universality across all targets. Furthermore, we show that the same optimal rate is achieved by performing the fine-tuning step via one-pass stochastic gradient descent. We support our theoretical analysis via experiments on both synthetic settings and language models.

0.19

Model Capacity Frontier: A Development Framework for LLM-Supported Social Simulation

cs.MA cs.AIcs.CY ·2026-10-16 ·Patrick Zhao, Yuxuan Li, Hirokazu Shirado
Abstract

Large language model (LLM) agents are increasingly incorporated into agent-based models (ABMs) on the premise that richer, context-sensitive human behavior can improve empirical fidelity. However, LLM-supported ABMs introduce an open-ended design space in which subtle modeling and configuration choices, such as prompt design and environment coupling, can substantially alter outcomes. Developing LLM-supported ABMs to simulate human collective behavior therefore requires navigating configuration sensitivity while identifying the limitations of the current model architecture. In this paper, we introduce the Model Capacity Frontier (MCF), a framework that characterizes the empirical performance attainable by a specified architecture across its admissible configurations. In multi-objective settings, this capacity is represented by an estimated Pareto frontier obtained through systematic configuration exploration. We demonstrate MCF in crowd simulation of university commencement events with approximately 12,000 agents, using empirical movement data on temporal occupancy, spatial distribution, and flow direction. We first characterize the capacity of a conventional crowd model and then evaluate successive LLM-supported architectural extensions. In the process, MCF reveals which dimensions of attainable performance expand and which remain limited, providing empirical targets for refinement and re-evaluation. Initial LLM integration improves flow direction but degrades temporal occupancy and spatial distribution; subsequent architectural refinement of the LLM layer improves temporal occupancy and recovers spatial performance, while retaining much of the earlier gain in flow direction. Our case study shows how MCF guides the systematic architectural development of LLM-supported ABMs in configuration-sensitive, open-ended design spaces.

0.19

Quality-of-Transmission Estimation in Quantum Networks: Compute, Simulate, or Learn?

cs.LG quant-phcs.NI ·2026-10-16 ·Anis Bekri, Amar Abane, Van Sy Mai, Abdella Battou, Saddek Bensalem
Abstract

Future quantum networks will rely on control-plane decisions such as routing, scheduling, entanglement provisioning, purification, and memory allocation. Each decision depends on a timely and reliable estimate of the quality of the entanglement that a candidate path can deliver, with end-to-end fidelity as a central metric. Inspired by the optical-networking experience with quality-of-transmission (QoT) estimation, this article asks when a quantum controller should compute fidelity analytically, simulate it, or learn a surrogate model. The optical analogy is useful but incomplete: optical QoT estimation often predicts a scalar transmission margin, whereas quantum QoT estimation must account for probabilistic entanglement generation, random memory waiting times, decoherence, state consumption, and measurement constraints that can make fidelity a stochastic target. We propose a compute/simulate/learn framework for quantum QoT estimation and illustrate it with a simulation-based proof of concept. Our study shows that analytical composition can be sufficient in simple regimes, that simulation becomes necessary when stochastic memory effects shape the delivered fidelity distribution, and that lightweight learned surrogates can approximate simulator labels at much lower runtime cost for online control. We close with open challenges for deployment, including stochastic targets, sim-to-real transfer, data-generation cost, physics-informed learning, and generalization beyond individual paths to full network topologies.

0.18

When to Forget: Risk-Controlled Context Adaptation for Long-Horizon Agents

cs.AI cs.LG ·2026-10-16 ·Chenle Chen, Yangbo Wei, Zhijie Zheng, Zhenyu Su, Qifan Wang +4 more
Abstract · 32 pages, 5 figures, 14 tables

Long-horizon agents learn from failure, but the value of a failure record can change as execution progresses. We study when retaining, compressing, or removing the same history block improves subsequent execution relative to full-history KEEP. We introduce RECAP, a framework that combines matched context interventions, execution-relational prediction, and independent policy certification. Matched environment states and rollout seeds measure success and continuation prompt-cost differences; prefix-only relations between a failure block and the current execution propose promising edits. Independent task bundles then select a policy subject to harmful-intervention and non-positive-benefit risk constraints under a fixed measurement protocol. A five-seed longitudinal case holds both the failure block and its summary fixed: compression fails at an early state and becomes a cost-saving intervention at a later state. Across five backbones, RECAP produces positive paired-success saving estimates with backbone-specific reliability-efficiency tradeoffs. On Hunyuan-A13B, it edits 43.24% of retained checkpoints while matching KEEP's observed 91.89% continuation success and reducing unconditional mean continuation prompt cost by 4,619.8 tokens (2.13%). The framework provides an execution-level reference for deciding when to forget: retain history by its value to what comes next.

0.18

Distill Until Divergence: Error-Truncated Hindsight Distillation for Agentic RL

cs.AI ·2026-10-16 ·Yibin Huang, Conghui Zhu
Abstract

Long-horizon interaction requires agents to make multiple decisions in a changing environment. Outcome-based GRPO lacks fine-grained guidance for intermediate decisions. Recent methods distill completed trajectories into reusable skills and use hindsight-conditioned OPD to provide dense token-level credit. However, skill-induced probability shifts do not necessarily represent reward-aligned credit; especially after an erroneous action changes subsequent interaction states, skill-conditioned supervision may become mismatched with the current state. To address this mismatch, we propose \methodname{}, which routes two complementary hindsight skills according to trajectory outcomes and error boundaries: Positive skills capture successful workflows to reinforce successful trajectories and pre-error decisions, while Negative skills capture failure-specific corrections to target decisive errors. The post-error suffix is optimized exclusively by GRPO, retaining outcome-based learning without extending potentially mismatched hindsight supervision. Extensive experiments demonstrate competitive task performance and support effective fine-grained credit allocation.

0.18

LangNCO: Compiling Natural-Language Constraints into Inference-Time Interventions for Neural Combinatorial Optimization

cs.LG cs.AI ·2026-10-16 ·Hironori Ohigashi
Abstract

Neural combinatorial optimization (NCO) rapidly produces high-quality solutions for problem structures encountered during training, but operational requirements may change after deployment and be communicated in natural language. We propose LangNCO, a framework that compiles instance-specific requirements into structured inference-time interventions for a frozen NCO solver, without changing the pretrained policy's architecture, learned input representation, or parameters. A pretrained LLM serves as a replaceable semantic front end, translating each requirement into a schema-validated domain-specific language (DSL) rule set. A deterministic Rule Executor applies the rules as nonexpansive hard action masks. Because local masking prevents immediate violations but does not ensure that a complete feasible continuation remains reachable, Critical-Offset Decoding extracts rule-referenced critical nodes and reallocates part of a fixed rollout budget to multi-strength biased rollouts. Using controlled task-specific language to isolate solver adaptation from open-ended language robustness, we evaluate a CVRP policy on road closures, time windows, vehicle assignments, visit-order requirements, all pairwise combinations, and twelve additional constraint types. LangNCO achieves 100% feasibility on the four primary datasets with mean reference gaps of 1.63--3.26%, 97.33--100% feasibility when separately generated rules are composed for pairwise constraints, and 98.67% feasibility across the additional constraints. These results demonstrate modular post-training adaptation of a frozen NCO solver through inspectable language-derived rules, without retraining.

0.18

What Remains Without Leaking Test Labels? Exposing Evaluation Flaws in Graph-Level Anomaly Detection

cs.LG cs.SI ·2026-10-16 ·Yujie Lu, Da Yan
benchmark
Abstract · v2: Retitled. Adds a replication in UB-GOLD's Type III scenario, two extra training seeds, and three analyses of where the checkpoint-select…

Graph-level anomaly detection (GLAD) identifies unusual graphs in applications such as molecular screening, often using only normal graphs for training. Yet the evaluations of some detectors select models with test labels, and high benchmark scores alone do not show whether detectors recognize structural changes. We audit released code and compare the area under the receiver operating characteristic curve (AUROC) for model states (checkpoints) chosen with and without test labels from the same training run. We separately compare test-tuned hyperparameters with repository defaults and test sensitivity to structural changes while controlling graph size. We evaluate 21 detectors on the UB-GOLD benchmark and on synthetic and molecular graphs with controlled structural changes. About half of the checkpoint-selection gain is linked to choosing a checkpoint and reporting its performance on the same labeled test set. The mean gap persists across training seeds and when normal and anomalous graphs come from different datasets. Most detectors miss many tested structural changes that simple counts detect. With tuned hyperparameters, the two leaders remain the top two by AUROC under every checkpoint rule. Across nine main datasets, test-label checkpoint selection raises the mean AUROC of UB-GOLD's 12 detectors by 5.12 points relative to the final checkpoint.

0.18

Cache2LoRA: Understanding and Improving Context Internalization

cs.AI cs.CLcs.IR ·2026-10-16 ·Yi Fang, Gaurav Srivastava, Zhenyu Bi, Sirui Ding, Xuan Wang
Abstract

Context-to-LoRA methods encode documents as low-rank parameter updates, allowing language models to answer without retaining the source text in their decoding context. How reliably do these updates preserve facts and support their use under different instructions? Controlled diagnosis of existing context-to-LoRA methods separates fact discrimination, target retention, and response adaptation. Competing sources influence answers, while target-only construction and memory-state optimization improve recovery at fixed update size. Recovering facts also does not ensure that answers follow response instructions. We introduce Cache2LoRA, which uses each query to aggregate evidence from a reusable frozen document cache before generating LoRA updates. Trainable encoding operates only on the query and memory-token suffix, and multi-style supervision teaches both evidence preservation and answer expression. Across 17 tasks, Cache2LoRA improves on existing context-to-LoRA methods and achieves macro F1 of 0.663, approaching full-context in-context learning at 0.671. For 32K documents and 256-token outputs at batch size 16, its decoding throughput is 5.9$\times$ that of full-context ICL. At matched memory size, generated LoRA preserves answer accuracy more reliably than the learned-KV variant under long, diverse intervening text and combines more effectively with retained document KV, yielding higher hybrid F1 and EM.

0.18

Continuous Ontology Assurance through External Resources -- An Ontology Engineering Method with a Case Study in Software Language Engineering

cs.AI ·2026-10-16 ·Ralf Lämmel
Abstract · Under review

Domain ontologies require review against knowledge recorded in evolving external resources. We propose COATER, a method that manages executable review components called arrows. Each arrow compares a scoped part of an ontology with a relevant resource and reports support or challenges to its axioms. Challenges may prompt proposals for completion, correction, or restructuring and become tasks for human or agent-assisted triage. Per-axiom coverage summarizes the assessments. We illustrate COATER with the FSL ontology in software language engineering (SLE): arrows review language coverage, language conceptualization, and software-engineering activities. An LLM-assisted linkage exercise increases compliance with a defined Wikipedia-linkage policy from 405 to 600 of 600 entities. A structured catalogue of SLE resources organizes review opportunities.

0.18

Alliance of Stochastic Routing and Reinforcement Fusion for Multi-Model Reasoning

cs.LG cs.MA ·2026-10-16 ·Selim Furkan Tekin, Ling Liu
theory
Abstract

This paper studies dynamic model routing under uncertainty in multimodal settings, where a system must select and potentially compose multiple models to answer a query. Standard reinforcement learning approaches for routing often suffer from premature convergence to suboptimal models due to noisy rewards and insufficient exploration. We propose a unified vision-language framework (VL-Scout) for efficient multimodal and multi-step reasoning with turn-based reinforcement fusion. First, VL-Scout combines Bayesian priors, policy gradient optimization, and sequential decision-making to enable structured exploration and adaptive inference. A context-dependent Dirichlet belief model allows for epistemic uncertainty over model quality and is updated online based on observed rewards and context-specific evidence, enabling Thompson-style exploration while serving as a prior for regularization. Second, a routing policy is trained via a KL-regularized REINFORCE objective by maximizing a reward-weighted evidence lower bound, where the policy acts as an amortized posterior, and the Dirichlet belief serves as a prior. Third, we introduce a fusion policy trained with Group Relative Policy Optimization (GRPO) or Reinforcement Learning with Verifiable Rewards (RLVR), which operates over intermediate model rationales and selects between two actions: CONTINUE or ANSWER, terminating with a generated output. We evaluate VL-Scout on four representative multi-modal benchmarks (MMMU, MMMU-Pro, OKVQA, and RealWorldQA), compared against 11 representative routing and policy optimization baselines. Our method provides consistent performance improvements while reducing premature policy collapse. These results highlight the importance of integrating uncertainty-aware priors with both amortized policy learning and sequential fusion for robust and efficient model selection in discrete action spaces

0.17

Residual Stochastic Concept Bottleneck Models

cs.LG ·2026-10-16 ·Stephen Marcou, Mateo Espinosa Zarlenga, Sonia Laguna, Moritz Vandenhirtz, Julia E. Vogt +1 more
Abstract

Concept-based models (CMs) learn human-interpretable intermediate representations that enable users to understand the model's reasoning and steer its predictions through test-time concept interventions. However, in realistic settings, their performance can be limited by concept incompleteness, where the training concepts fail to capture all task-relevant information. Adding an unconstrained residual pathway to the model's predictive process can overcome the consequences of incompleteness. However, this introduces an inscrutable, leak-prone prediction pathway that may weaken concept interventions, as these often leave residual variables unchanged. To address this limitation, we introduce Residual Stochastic Concept Bottleneck Models (Res-SCBMs), a family of concept-based models that jointly model concept and residual variables through a multivariate Gaussian distribution. Res-SCBMs learn a covariance structure that explicitly captures dependencies between the known concepts and learnt residual variables, allowing interventions to propagate to statistically dependent residual variables through Gaussian conditioning. This extends the intervention capabilities of previous CMs without requiring concept completeness during training. Our results show that, across multiple concept-incomplete benchmarks, Res-SCBMs improve predictive performance over competing baselines while retaining effective test-time concept interventions.

0.17

A Mixture-Gated Theory of Anisotropic Representation Dynamics in Transformers

cs.LG ·2026-10-16 ·Nguyen Do, My T. Thai
Abstract · 62 pages

Small differences between internal representations can shrink or grow as they pass through a Transformer layer, affecting how information is preserved and how sensitive the model is to perturbations. Yet it remains unclear when changes in attention weights can turn a representation difference that would otherwise shrink into one that grows. We develop a Mixture-Gated Theory of Anisotropic Representation Dynamics that separates self-attention into value transport, query gating, and key gating. Our analysis reveals how the geometry of attended representations constrains gating and how interactions between these mechanisms can reverse contraction induced by value transport. We formalize this phenomenon as gating-induced regime switching, where a representation difference contracts under fixed attention routing but expands under full attention. We derive a necessary and sufficient spectral criterion that certifies the existence or absence of switching and identifies switching directions without exhaustive search. Beyond existence, our theory characterizes maximum relative switching strength, certifies the severity of contraction that gating can overcome, and determines the largest dimension of a subspace in which every nonzero direction switches. Experiments across four Transformer architectures validate these predictions through analytical and finite-forward comparisons, revealing distinct and joint query--key gating mechanisms and nontrivial switching subspaces. Our results provide a mechanistic and spectral understanding of how attention routing can alter local representation dynamics.

0.17

Learning without relearning physical symmetry: an SO(3)-equivariant convolutional network (CNN) on the sphere

cs.LG stat.MLastro-ph.CO ·2026-10-16 ·Joshua Williamson, Niall Jeffrey, Lorne Whiteway
Abstract

Many of the most important datasets in the physical sciences live natively on the sphere, and these data often carry no preferred position or orientation. Using methods that do not reflect this symmetry leads to far higher simulation requirements, which can be catastrophic for scientific analyses (forcing reliance on cheap, imperfect models that bias results). We present a novel spherical implementation of a gauge equivariant mesh architecture that is specifically designed for the Hierarchical Equal Area isoLatitude Pixelization (\healpix{}) of the sphere. The $\mathrm{SO}(3)$-equivariant convolution is built from steerable filters, parallel transport and spin-weighted features. The architecture supplies the exact symmetry, so the network does not spend training data learning it. The convolution operations are constructed to give provably exact equivariance, given a band-limit. Our full CNNs can achieve SO(3)-invariance error at per-cent-level and lower, which is controlled through well-understood hyperparameters. On benchmarks, our technique gives the most accurate dense gradient regression and competitive cosmic microwave background foreground removal and denoising. The improved simulation efficiency matters most in simulation-based inference (SBI), where high-fidelity simulations are expensive and neural compressors require large training sets, so a fixed budget forces analyses with less accurate simulators. In cosmology lensing SBI, our technique yields the tightest posteriors at matched simulation count. This approach is a novel method that will enable new convolutional neural network analyses, including generative and inference tasks, for processes and data on the sphere.

0.17

Certified neural approximations for parabolic obstacle problems

cs.LG math.NAmath.FA ·2026-10-16 ·Davood Damircheli
theory
Abstract

Residual-trained neural surrogates carry no error guarantee, a limitation for free-boundary problems such as phase change and contact. We derive guaranteed error bounds for neural approximations of parabolic obstacle problems in first-order state--flux--multiplier form. The main result bounds the flux error of the network and the energy error of its feasible projection $\max\{v,g\}$ by the constitutive residual, the initial mismatch, the complementarity defect, and the PDE residual weighted by $c_F^2/α$, without the coercivity and norm-equivalence constants of the first-order formulation. It extends the Apushkinskaya--Repin majorant for the heat operator to general coefficients, free multipliers and a flux bound. A penetration-consistency result shows that neural triples with arbitrarily small bound exist. Outward-rounded cellwise Taylor-model enclosures turn the bound into a finite computation valid on any adaptive partition. The flux and gradient errors of the projected output are certified below $0.27$ and $0.33$ of the solution scale for a one-dimensional Stefan problem and, after adaptive refinement, below one half for all three trained networks on a two-dimensional expanding disk. On an expanding ellipse the gradient bound stays above one half; substitution diagnostics trace the floor of the trained networks to the projected state near the free boundary, and the remaining gap to enclosure overhead. Without a closed-form solution, a validated lower bound replaces the solution norm; there the enclosure is nearly exact, but the trained networks have errors above the solution scale and no useful accuracy is certified. The bounds control a state--flux norm, not the multiplier or the free-boundary location.

0.17

Monitor-Validity Failures: LLM Safety Evidence Paths at Evolving Framework Boundaries

cs.AI cs.CRcs.SE ·2026-10-16 ·Junchi Liao
Abstract

Modern LLM applications rely on runtime monitors that inspect tool calls, traces, streams, detector outputs, and verifier inputs. This creates an interface-reliability problem: a framework path may remain operational while policy-relevant evidence never reaches its decision point. In one representative case, a regex detector returned a positive match, but the framework stranded the result before the block decision. This differs from jailbreaks and detector errors because the evidence is already recognized yet disconnected from enforcement. We study this repair-boundary signature with canary contracts: parent/fixed confirmation units that place a known payload only on the changed evidence surface and record monitor visibility together with the downstream decision for action-boundary cases. Applying the protocol to public framework repairs, we confirm 18 evidence-path failures across 11 LLM or agent safety project groups and four implementation ecosystems. Nine are established-boundary repairs, eight are new-surface policy-coverage gaps, and one is a boundary-only evidence case. Fourteen reach E1 enforcement boundaries: seven native action outcomes, six configured enforcement boundaries, and one synthetic-consumer consequence. The remaining cases affect verifier, safety-prompt, or observability channels. The cases form carrier, adapter, authority, and surface families and show how ordinary application behavior can remain intact while a safety-specific evidence path misses its decision. Exact revisions, deterministic outcomes, negative controls, and a frozen prospective audit document the confirmation rule. Monitor assumptions therefore need regression tests at the schemas, streams, tools, and adapters that frameworks evolve.

0.17

TriRouter: Sparse-to-Dense Electrode-Space Routing for Robust EEG Motor Imagery Decoding

cs.LG eess.SP ·2026-10-16 ·Ying Zhou, Yaoxin, Mengting Liu
Abstract · Submitted to IEEE ICASSP 2027

Motor imagery brain-computer interfaces require reliable decoding from multichannel EEG, but practical acquisition often yields sparse or corrupted electrode montages. We propose TriRouter, an electrode-space routing framework with discriminative-preservation, mask-robustness, and latent-compensation paths. A validation-calibrated router uses observable channel availability to allocate path contributions, preserving complete-channel discrimination while increasing representation recovery under severe missingness. In BNCI2014-001, TriRouter outperforms standard public backbones and a dynamic-spatial-filtering corrupted-EEG baseline. It improves accuracy over the strongest corrupted-EEG baseline by 3.68, 4.02, and 6.46 percentage points at nominal 25%, 50%, and 75% missing settings, and remains effective under structured regional missingness and additional low-channel dataset evaluation.

0.17

From Memory Hits to Answerable Evidence: Query-Conditioned Memory Compilation for Long-Term Agent Memory

cs.AI ·2026-10-16 ·Shangze Li, Yanyu Yao, Derong Xu, Zhi Zheng, Tong Xu
Abstract

Long-term agent memory requires recovering answer-relevant evidence dispersed across turns and sessions. Existing systems increasingly use structured memories and adaptive retrieval, yet often pass retrieved or expanded units directly to the reader, coupling memory addressing with evidence presentation and leaving the final context redundant or incomplete. To address this gap, we propose MemCompile, a query-conditioned evidence compilation framework for long-term conversational memory. Given a query, the Address Plane locates relevant support through compact source-linked descriptors and capacity-aware Cells, while a normalization boundary resolves heterogeneous retrieval hits to stable SourcePacket identities. Its Evidence Plane then allocates a bounded evidence budget across complementary packets, admits limited additional support, and renders the retained claims associated with each selected packet into an answer-ready context. MemCompile achieves 80.52% Judge Accuracy on LoCoMo and extends the empirical quality-cost Pareto frontier to a higher-quality regime. Its strong performance on LongMemEval-S further suggests that the benefits of query-conditioned compilation generalize across long-term conversational-memory benchmarks.

0.17

Generative Policy Improvement from Observational Data

cs.AI cs.LG ·2026-10-16 ·Jonas Schweisthal, Dennis Frauen, Stefan Feuerriegel, Emma Brunskill
Abstract · Preprint

We study how to improve decision policies from observational data to generate better actions when actions are complex, structured objects, such as text messages or images. This task is subject to two major challenges: (i) action-level overlap is often too weak for reliable off-policy evaluation and improvement, and (ii) generating new structured actions often requires expensive training or fine-tuning (e.g., of an LLM) and can be ill-defined due to information loss in action representations. To address both, we propose PolicyLift, a three-stage framework for generative policy improvement with structured actions from observational data. In Stage 1, PolicyLift learns a small set of discrete action codes that capture high-level strategies in the observed action space and are tailored to support policy improvement under limited overlap. In Stage 2, it learns a context-dependent policy over these action codes. At deployment time, Stage 3 "lifts" the learned code-level policy back to the original action space by generating actions from the historical behavior system in a code-conditioned way. We show theoretically that improving the policy over action codes guarantees improvement of the induced policy in the original action space. We demonstrate PolicyLift across experiments in synthetic and real-world text generation tasks with consistent policy improvements over the behavior policy.

0.17

Trapped in Your Story: Evaluating Narrative-Induced Sycophancy in Large Language Models

cs.AI ·2026-10-16 ·Runtai Yang, Dandan Zhu, Kaiwei Zhang, Guangtao Zhai
benchmark
Abstract

Large language models (LLMs) are known to exhibit sycophancy by adjusting their responses toward users' expressed views. However, real users often justify their opinions through reasons, personal experiences, causal explanations, and broader narratives, raising the question of whether such narratives can further steer model judgments. We construct a controlled benchmark of 500 debatable questions across ten domains, where the user stance is held fixed while its justification is systematically varied without introducing new task-relevant evidence. Each question is instantiated under 11 matched conditions: one neutral baseline and five mirrored narrative levels for each of two opposing positions, yielding 5,500 evaluation instances. We evaluate seven LLMs and observe substantial differences in narrative-induced agreement. The effect is not monotonic with narrative elaboration: personal experience can reduce agreement, whereas causal framing and worldview-based synthesis increase it. Controlled GPT-5.1 ablations further show that causal framing produces the strongest positive effect, while claim-preserving bullet presentation also increases agreement over paragraph form. These findings suggest that sycophancy is shaped not only by user stance, but also by how supporting explanations are structured and presented.

0.17

LexPilot: Benchmarking the Proactive Interaction Ability of Legal LLMs with Laypeople in Chinese Civil Case Consultation

cs.AI ·2026-10-16 ·Wenbin Zhang, ShangZhan Li, Bingxiang He, Ye He, Yuxin Zhou +3 more
benchmark
Abstract

Large Language Models (LLMs) have greatly assisted legal professionals. However, our empirical studies reveal that, when faced with underspecified consultations from non-experts, current LLMs often produce irresponsible responses with limited utility despite lacking sufficient case details for definitive conclusions. To better assist laypeople, legal LLMs must identify focal points of dispute and proactively elicit crucial context from users. Unfortunately, existing data resources often focus on interaction and response quality, offering limited insight into how proactive interaction affects LLMs' downstream claim-level judgment prediction in actual civil cases. To address this gap, we introduce \textbf{LexPilot}, a benchmark linking proactive legal interaction to claim-level verdict prediction using authentic Chinese civil judgments. Constructed from 432 real-world judgments across 36 common causes of action, LexPilot comprises expert-reviewed pairs of complete and underspecified consultation histories with corresponding court verdicts. Experiments reveal that current LLMs struggle with proactive legal interactions in underspecified consultations, with downstream judgment prediction substantially underperforming that in the fully specified consultation setup (-23.84% on average). Further analysis identifies critical bottlenecks hindering these capabilities. Our work highlights the urgent need for more user-centered legal AI to better serve the public.

0.16

Large Diffusion Models Need More Than EMA

cs.LG ·2026-10-16 ·Chuyu Hu, Huanran Chen, Shitong Shao, Yinpeng Dong
Abstract

Large-scale diffusion models are commonly trained with a constant learning rate (LR) and exponential moving average (EMA), despite the strong performance of LR decay and Schedule-Free optimization in training large language models. We study whether EMA is uniquely suited to diffusion training, or whether stronger alternatives have been underexplored. Our theory shows that averaging a constant-step trajectory has an inherent optimization limitation, whereas LR decay and Schedule-Free optimization admit vanishing stationarity guarantees under standard nonconvex assumptions. Empirically, LR decay and EMA provide complementary benefits: combining decay with EMA further improves the standard baseline at most tested starting points, although the gain remains sensitive to when decay begins. Schedule-Free avoids this timing requirement and consistently outperforms EMA with a constant LR across training stages, achieving both lower fixed-evaluation objectives and better generation quality. Our analysis further shows that all three strategies substantially reduce parameter variability, yet stronger stabilization is not always better: EMA with a constant LR retains comparatively higher parameter variability, while aggressive decay combined with EMA achieves the lowest checkpoint variability but also weakens continued optimization. Schedule-Free substantially reduces parameter variability while preserving more late-stage progress, providing a better balance between parameter stability and continued optimization. Finally, the benefits of Schedule-Free extend beyond pretraining to ReFL and generator-side DMD, demonstrating its transferability across distinct diffusion optimization settings. Overall, our results suggest that large diffusion models need more than parameter averaging alone: effective optimization requires balancing variance reduction with continued optimization.

0.16

SkillBrowser: Selective Skill Exposure for Scalable Skill-Using Agents

cs.AI ·2026-10-16 ·Jin Ye, Yifan Jiang, Lan-zhe Guo
Abstract · 23 pages, 3 figures. Submitted to ACL Rolling Review, October 2026

Large language model agents increasingly rely on skills to extend their capabilities in complex tasks. However, scalable skill use requires Selective Skill Exposure (SSE): providing relevant procedural content when needed while avoiding unnecessary exposure that may disrupt reasoning. Existing paradigms fall short in complementary ways. Progressive disclosure exposes metadata for a growing skill library, introducing context pollution; retrieval limits catalog exposure, but static retrieval from the initial task query may be misaligned with capability needs that emerge during reasoning and may still inject unnecessary skills. We propose SkillBrowser, a Meta-Skill-managed framework that enables an agent to identify current capability needs, formulate targeted queries, and selectively load skills on demand. Across six SRA-Bench subsets and five LLMs, SkillBrowser improves aggregate accuracy over static retrieval and LLM-based selection, including multi-step tool use, and mitigates accuracy degradation on already-solvable instances. Benefits remain model- and task-dependent, with under-invocation of useful skills and additional latency as key limitations. These results demonstrate the viability of Meta-Skill-managed SSE for scalable skill repositories.

0.16

RTSM: Adapting Recursive Models to Multivariate Time-Series Classification

cs.LG ·2026-10-16 ·Georg Tenzing, Anna Tegon, Luca Benini, Thorir Mar Ingolfsson
Abstract · 33 pages, 6 figures

Accurate multivariate time-series classification requires extracting temporal features and integrating information across a sequence. Weight-tied refinement offers a way to repeatedly process these features without allocating new parameters at each stage. Applying this principle to time series requires choosing both a temporal representation and a latent-state organization suited to classification. We introduce the Recursive Time-Series Model (RTSM), which combines a learned convolutional temporal tokenizer, single-state weight-tied refinement, and a pooled classification readout. Across all 30 UEA datasets, RTSM achieves 72.81% mean accuracy under validation-based configuration and checkpoint selection, the highest observed mean among the 17 evaluated deep-learning models and 1.60 percentage points above MambaSL. Ablations on eight datasets favor the proposed tokenizer over a simple convolutional embedding and the single-state model over the tested two-state variants. The complete classifier reaches 86.10% mean accuracy versus 84.79% for a separately trained tokenizer-only classifier. These findings identify temporal encoding and state organization as key design choices for adapting recursive architectures to time series. The resulting classifier combines competitive accuracy with a shared refinement module whose parameter count is independent of computational depth.

0.16

CoTA++: Understanding the Repeat Curse in dMLLMs from an Information Flow Perspective

cs.AI ·2026-10-16 ·Qiyan Zhao, Xiaofeng Zhang, Xu-Yao Zhang, Shijie Han, Zhepeng Li +2 more
Abstract

Diffusion-based multimodal large language models (dMLLMs) decode by iterative denoising with bidirectional attention, which precludes the standard autoregressive key--value cache. Practical deployments therefore rely on approximate caching that reuses intermediate states across denoising steps. Such caching, however, often makes a response repeat one token at adjacent positions in a run that breaks the sentence, a failure we term the \emph{Repeat Curse}. Tracing the context-token information flow reveals three disruptions. Within a step, caching disperses the attention that anchors a position on nearby context tokens. Across steps, similarity-based refresh is not staleness-aware, so some positions are decoded against stale key--value states. Across layers, the deep-layer entropy of the context fails to converge where repetition arises. We accordingly propose CoTA++, a training-free, plug-and-play framework that counteracts each disruption where it originates: Context-Token Attention Re-anchoring re-forms the attention routing against the current context, Decode-Aware Refresh reserves part of the cache's refresh budget for the positions about to be committed, and Context-Token Entropy-Guided Voting discounts the confidence of a position by the deep-layer entropy of its context, which reorders the unmasking. The three components act at non-overlapping stages of a decoding step and compose freely with existing dMLLMs and caches. Across three dMLLMs and two caching backends, CoTA++ reduces the sample repetition rate of long responses from up to 99.6\% to at most 17.0\%, recovers most of the accuracy lost to caching on nine benchmarks, and keeps the speedup of caching at under 15\% overhead, making approximate caching reliable for dMLLMs.

0.16

Universal Semantic Enhancement for Weakly Supervised Anomaly Detection

cs.LG ·2026-10-16 ·Siyu Zhou, Yuqin Xie, Hao Yuan, Yingjie Zhou
code/data
Abstract · https://github.com/Simon-zsy/USE-AD

Weakly supervised anomaly detection on tabular data is an essential task in many label-scarce applications, since tabular data is widely encountered and it is commonly observed in practice that only a limited number of labeled anomalies are available alongside abundant unlabeled samples. Most existing methods try to learn features that capture the inherent distribution patterns in the target domain. However, such features often struggle to provide a complete description of possible abnormal characteristics and are highly susceptible to overfitting the extremely scarce labels. To address this challenge, we propose a weakly supervised anomaly detection method based on universal semantic enhancement for tabular data, which fuses the universal knowledge from a pre-trained tabular foundation model with the domain-specific knowledge learned on the target dataset. The two types of knowledge are naturally complementary: universal semantics describe the common patterns of tabular data across domains, while the domain-specific features capture the inherent distributions of normal and anomalous behaviors in the target dataset. A prototype-conditioned modulation mechanism is then designed to effectively integrate the two types of features, allowing them to reinforce each other for improved anomaly detection. Extensive experiments on public datasets show that the proposed method outperforms competitive methods when only eight labeled anomalies are available in the training procedure.

0.16

RECREG: Recovery-based Shielding for Regular Safety Properties in Reinforcement Learning

cs.LG cs.AI ·2026-10-16 ·Alexander W. Goodall, Francesco Belardinelli
Abstract

In many real-world scenarios we need to consider formal requirements, such as safety and constraint compliance, to deploy Reinforcement Learning (RL) agents. In this paper we introduce RecReg, a safe RL framework capable of enforcing non-Markovian regular safety properties via recovery-based shielding. RecReg is a unified approach that supports different levels of a priori knowledge and safety guarantees. The paper primarily focuses on building the theoretical foundations for RecReg in the tabular setting, providing rigorous safety and convergence guarantees that are not present in related work. We provide an extensive set of empirical results using the MASA-Safe-RL library, and we outline an extension of RecReg to deep RL domains using SOTA model-based RL. Our empirical results demonstrate that RecReg achieves efficient learning and matches the theory in this paper.

0.16

Conformal Policy Selection: When to Deploy AI Decisions for Targeted Intervention

cs.LG stat.MEstat.ML ·2026-10-16 ·Junyu Cao, Ruijiang Gao, Maytal Saar-Tsechansky, Mingzhang Yin
theory
Abstract · 28 pages, 6 figures, 4 tables

Artificial-intelligence (AI) policies are increasingly used to make consequential decisions in medicine, finance, education, and public services, yet deploying them responsibly requires knowing not only whether a policy helps on average, but for whom. Existing tools for off-policy evaluation and safe policy learning provide population-level guarantees, but some individuals may experience adverse treatment effects. We recast targeted deployment as a multiple-testing problem: for each candidate unit, test whether the outcome under the AI policy would exceed a user-specified threshold, and control the false discovery rate (FDR) among the selected set. Our method, Conformal Policy Selection, combines weighted conformal $p$-values that account for the distributional shift from behavior to target policy with a two-step selection-and-pruning procedure. We prove finite-sample FDR control and establish a doubly robust asymptotic guarantee, where FDR remains controlled if either the outcome model or the propensity model is correctly specified. Synthetic experiments verify these guarantees across four misspecification regimes, and a semi-synthetic study built on real electronic-health-record covariates illustrates how the same FDR control can act as a deployment safety gate, bounding the harmful treatments that a globally trained policy would otherwise administer to a high-risk subpopulation where it misgeneralizes. The framework offers a principled method for safe AI deployment in high-stakes applications.

0.16

Task-aligned spectral inductive biases in parameterized quantum circuits

cs.LG quant-ph ·2026-10-16 ·Felix Frohnert, Adrián Pérez-Salinas, Sebastian Lange, Marcus Schwaebisch, Evert van Nieuwenburg +1 more
Abstract

Parameterized quantum circuits are widely investigated in quantum machine learning, yet how their architecture can provide task-aligned inductive biases remains poorly understood. We study data reuploading quantum circuits with two distinct inputs: a signal that varies between samples and a fixed, task-defined spectral query that specifies how the signal should be probed. We show that encoding the query can change the circuit's hypothesis class over the signal by modulating how its existing signal frequencies are combined, without introducing new sample-dependent information. This mechanism matches tasks built on nonuniform Fourier transforms, whose evaluation frequencies specify the query and whose sampling positions guide its encoding. We examine this connection in a classification task inspired by synthetic aperture radar, where acquisition geometry determines a query for classifying unfocused spectral patches before image formation. Across small-scale spectral and radar-like tasks, aligned queries improve learning from limited data over parameter-matched circuits with mismatched or absent queries. These results provide a route from known spectral task structure to the design and validation of inductive biases in parameterized quantum circuits.

0.15

Agentic In-Context Learning: What Algorithms Can Transformers Learn in Closed-Loop Interaction?

cs.AI ·2026-10-16 ·Shiguang Wu, Yaqing Wang, Quanming Yao
Abstract

The learning algorithm view of in-context learning (ICL) studies how fixed-parameter Transformers learn from supplied examples. Language model agents extend this setting through interaction: their actions determine what evidence enters context next. We call this regime Agentic In-Context Learning (Agentic ICL). Existing studies largely depend on externally supplied contexts or reward-bearing interaction histories, while work on general tool-using agents lacks a learning algorithm view. We represent an interactive learner by its state, update, and decision rule, covering active learning, bandits, and online reinforcement learning (RL). For expressivity, we construct causal Transformers that approximate these algorithms over a finite horizon, and bound the resulting loss in return. For training, we characterize when optimizing process rewards preserves behavior that maximizes the final outcome. Distillation experiments show that Transformers can learn to execute three representative recursions. We then train models from outcome or process rewards without oracle action or state targets, as in RL post-training. Different rewards yield distinct decision policies. Improving the alignment of process supervision with outcome-optimal action preferences shifts learned decisions accordingly. Such behavioral differences persist in a pretrained language model across textual paraphrases.

0.15

Learning angles for quantum approximate optimization

cs.LG quant-ph ·2026-10-16 ·Johannes Jakubik, Isabelle Wittmann, Zaheed Gaffoor, Gciniwe Simphiwe Baloyi, Craig Mahlasi +3 more
code/data
Abstract

The Quantum Approximate Optimization Algorithm (QAOA) and its variants are leading algorithms to help solve challenging combinatorial optimization problems. QAOA maps these problems to Ising Hamiltonians such that their ground state is the desired solution. Candidate solutions are then drawn from a variational quantum circuit. The performance of QAOA hinges on a set of variational angles. However, finding good angles is an instance-specific and computationally expensive non-convex classical optimization task. Here, we learn the mapping from the problem instance to high-quality QAOA angles. Our learned models replace the per-instance optimization, that we call the solver, with a single forward pass that is several orders of magnitude faster than the solver with a near identical quality. Our models further generalize to unseen topologies such as Barabási-Albert and 2D grid graphs. Experiments on quantum hardware confirm that QAOA samples with angles produced by our open-source models are indistinguishable from solver angles. Crucially, models sensitive to the topology of the problem instance increase the quality of the QAOA angles. Our work amortizes a major bottleneck in QAOA by an up-front data generation and training cost. Importantly, this cost can be amortized over time in settings where many similar problems are optimized on a regular basis, as is often the case in applications for industry.

0.15

SpikeQ: Spike-Guided Post-Training Quantization for Spiking Neural Networks

cs.NE cs.AIcs.CV ·2026-10-16 ·Kai Sun, Peibo Duan, Zhanglu Yan, Guowei Zhang, Yongsheng Huang +1 more
Abstract

On neuromorphic chips, spiking neural networks (SNNs) consume far less energy than artificial neural networks (ANNs), but the limited memory of these chips requires low-bit synaptic weights and membrane potentials. Existing post-training quantization (PTQ) methods provide them without retraining and, like those for ANNs, minimize numerical errors such as the mean squared error (MSE). In an SNN, however, a quantization error reaches later layers only through the spikes it flips. We find that errors of equal MSE can cause very different accuracy losses, whereas errors that flip the same fraction of spikes cause the same loss. We therefore propose SpikeQ, a spike-guided PTQ framework. For weights, we select the level of each weight by simulating the downstream neurons and minimizing their spike errors. For membrane potentials, most existing methods quantize the state after the reset, whereas neuromorphic chips compare the threshold with the membrane potential stored at a fixed bit width. We target this pre-fire quantization and build a reset-aware grid that places the threshold on a rounding boundary, so rounding never flips the spike at the current time step. We evaluate SpikeQ on seven benchmarks of image classification, semantic segmentation, and object detection. With 2-bit pre-fire membrane potentials, every existing grid collapses, whereas SpikeQ retains most of the full-precision accuracy. With both at 4 bits, the models stay within 2.59 points of full precision, shrink their memory footprint 7.8--7.9 times, and use 6.4--6.9 times less energy.

0.15

From Trajectories to Differentiated Skills: Cognitive Differentiation for Self-Improving Agents

cs.AI ·2026-10-16 ·Yewei Zhao, Derong Xu, Wei Chen, Yin Xu, Peilun Zhou +4 more
Abstract · 22 pages, 7 figures, 17 tables, including appendices

Large language model (LLM) agents tackle complex tasks through reasoning and interaction with external environments. Agent skills provide persistent, transferable capabilities that guide task planning, inform contextual decisions, support tool use, and enable executable procedures. Yet existing skill-evolution methods largely organize knowledge around skill artifacts, overlooking the functional heterogeneity within individual skills and its implications for knowledge maintenance and use. Inspired by cognitive science, we propose Cognitive Differentiation in Skill Evolution (CDSE), which evolves a unified skill through four functionally distinct yet coordinated components: Meta, Experience, Tool, and Script. CDSE extracts evidence-grounded semantic gradients from successful and failed trajectories, consolidates them by functional role, and applies component-specific updates and validation with cross-component coordination and adaptive invocation. An analysis of 500 top-ranked skill packages shows that 96.8% contain knowledge spanning at least two functional roles. Across six benchmarks, three agent models, and two execution harnesses, CDSE achieves the highest scores in all 36 settings, improving over the Base agent by 7.6 percentage points and the strongest competing method in each setting by 2.6 points on average. Ablation studies further support the importance of functional differentiation and adaptive invocation.

0.15

Scalable Levenberg-Marquardt Training Framework for Deep Learning

cs.LG cs.DC ·2026-10-16 ·Xianghao Wang, Johan Barthélemy, John Taylor
Abstract

The Levenberg-Marquardt (LMA) algorithm is a widely used nonlinear least-squares solver known for its rapid convergence. For small-scale neural networks, LMA provides better convergence and architectural robustness than first-order optimizers like Adam. However, LMA imposes prohibitive memory and computational costs. These costs are primarily from the Jacobian matrix, whose dimensions scale with both the number of samples and parameters. This bottleneck prevents its use in large-scale neural networks. Recent efforts scale LMA using preconditioned conjugate gradient (PCG) solvers to avoid materializing the Jacobian. However, PCG only approximates the solution, requiring many network propagations and a task-specific preconditioner. Other studies optimize LMA to reduce costs, but they restrict their methods to specific tasks or network architectures. As a result, LMA remains impractical for training general large-scale neural networks. In this paper, we propose LeMA, a distributed LMA algorithm for large-scale neural network training on multi-GPU platforms. Our contributions are twofold: First, an exact dual formulation of LMA removes the memory dependence on the parameter dimension. Second, a Jacobian sharding strategy computes Jacobian shards on the fly and in parallel to avoid materializing the full Jacobian. For large-scale neural networks, LeMA reduces the memory footprint of LMA from quadratic in the number of parameters to quadratic in the batch size. This makes exact LMA steps feasible for billion-parameter models without approximate PCG solves. Furthermore, for a convolutional neural network with 1,000,022,632 parameters on an 8-GPU node, LeMA achieves a 7.97x speedup over a single GPU, corresponding to a 99.7% strong scaling efficiency.

0.15

ShadowFPT: Backdooring Federated Prompt Tuning via Shadow Triggers

cs.AI ·2026-10-16 ·Kun Zhai, Teng Li, Yunhao Feng, Xingjun Ma
Abstract

Federated Prompt Tuning (FPT) adapts large vision--language models by freezing the pretrained backbone and optimizing only lightweight prompt parameters across clients. Although this design improves communication and parameter efficiency, it creates an overlooked security risk. Since the backbone is shared and frozen, a malicious client can induce backdoor behavior through the representation space while keeping its uploaded prompt updates close to benign ones. We propose \textbf{ShadowFPT}, a targeted backdoor attack that exploits this frozen-backbone attack surface. ShadowFPT first pretrains a learnable \emph{Shadow Trigger} against the frozen CLIP visual encoder, using either auxiliary public data or the malicious client's local data, so that triggered inputs are steered toward the target class in representation space. During federated prompt tuning, the malicious client adapts the trigger under the current global prompt and then optimizes its local prompt on both clean and triggered samples. Only prompt parameters are uploaded to the server, while the trigger and frozen encoders remain local. By shifting most of the attack burden from prompt updates to trigger-induced representation steering, ShadowFPT achieves targeted misclassification while preserving prompt-space stealthiness. Across multiple datasets, aggregation rules, and non-IID federated partitions, ShadowFPT increases the attack success rate from 19.81\% to 90.36\% in our main setting, while maintaining clean accuracy. It remains effective across textual, visual, and joint vision--language prompt tuning. These results identify frozen backbones as stealthy and underexplored backdoor surfaces in federated prompt tuning, suggesting that defenses based only on prompt-update anomaly detection are insufficient.

0.15

Where Planning Changes the Outcome of Orbital Games

cs.MA cs.AIcs.ROeess.SY ·2026-10-16 ·Duncan Eddy, Mahdi Al-Husseini, Mykel J. Kochenderfer
Abstract · IEEE Aerospace Conference 2027

This paper introduces OrbitalGym, a simulation environment for studying when planning algorithms or system design determine the outcome of adversarial orbital encounters. In an orbital game, two non-cooperative teams of satellites maneuver during rendezvous and proximity operations to win a differential-reward objective. Advances in terrestrial game playing have prompted many methods for orbital games, but studies typically evaluate one method in one scenario, often against a passive or random defender. It is unclear whether better planning matters once vehicle capability (maneuverability, sensing, communications) and adversarial responses are considered. OrbitalGym is a JAX-based environment supporting varying numbers of agents, state representations (position, velocity, attitude, fuel), explicit or implicit communication, and proximity or conical field-of-view sensors. It compiles games to batched GPU code, enabling large-scale parallel simulation of policies across many configurations. This identifies regimes where policy choice affects outcomes and regimes where system capability dominates. We study two close-range games: pursuit-evasion, where one spacecraft tries to capture another, and lady-bandit-guard, where a guard prevents an attacker from reaching a protected asset. We compare PD, LQR, model predictive path integral control, and Monte Carlo tree search over ratios of thrust and propellant, initial separation, game duration, sensor capability, and the definition of capture. In pursuit-evasion the propellant ratio decides most encounters: against five of six evader policies, pursuer policy changes capture probability by 10 percent or more in at most 15 of 90 capability configurations. In lady-bandit-guard, policy matters more: against bandits able to reach the asset, a better guard policy raises defense probability by at least 10 percent in 56 to 74 of 90 configurations.

0.15

Unified Multiparadigm Fusion Intelligence for Rare Disease Diagnosis

cs.AI cs.LG ·2026-10-16 ·Jian Yang, Peiling Liu, Songda Li, Yu Cao, Yuqing Feng +8 more
Abstract · 26 pages, 5 figures

Rare diseases (RDs) affect an estimated 300 million people globally, leaving patients to navigate a protracted diagnostic odyssey constrained by clinical scarcity and fragmented medical knowledge. Artificial intelligence (AI) has progressively advanced RD diagnostics, from symbolic methods and deep learning to large language model (LLM)-based systems. However, each paradigm carries structural limitations: symbolic methods are bounded by knowledge completeness, neural models sacrifice interpretability, and LLM-based systems are vulnerable to evaluation biases like data leakage. In this study, we introduce FusionRare, a unified multiparadigm fusion framework integrating knowledge-driven symbolic reasoning, deep-learning pattern recognition, case-based patient retrieval, and LLM-based heuristic inference via statistical meta-aggregation and knowledge-enhanced agentic reflection. Evaluated on 2,528 strictly isolated clinical cases, FusionRare achieves state-of-the-art RD prioritization, attaining a macro-averaged Recall @1 of 51% for phenotype-only diagnosis and 79% for phenotype-genotype joint analysis. These results represent decisive advantages over the advanced agentic system DeepRare, frontier LLMs including GPT 5.5, and specialized tools like PhenoBrain and LIRICAL under stringent anti-leakage protocols. A randomized, crossover clinician-AI validation demonstrated that FusionRare elevates the true diagnostic rank in 45% of decisions, improves physician confidence in 62%, and reduces time to diagnosis by 21%. These findings indicate that reconciling these orthogonal AI paradigms significantly enhances diagnostic capabilities, bridging the gap between computational benchmarks and actionable clinical reliability. More broadly, this work provides a replicable framework for deploying trustworthy diagnostic AI across complex, high-stakes, and underserved medical domains.

0.14

GRAFT-EEG: Gauge-Aware Physical Field Transport for Cross-Dataset Motor-Imagery EEG

cs.LG eess.SP ·2026-10-16 ·Ying Zhou, Mengting Liu
Abstract · Under review at ICLR 2027

Cross-dataset EEG decoding often treats montage, reference, and active-sensor changes as representation shift, although they also alter the observation operator. We study target-statistics-blind transfer: fitting receives no target signal, label, or empirical statistic, and electrode coordinates only instantiate a fixed prediction operator. method removes the active-set common-reference gauge and transports each native mask through a template motor leadfield. A source-fitted spectral classifier anchors the decision; a bounded state-space-model (SSM) residual is secondary. We establish restricted identifiability and an operator-perturbation bound. Across five leave-one-dataset-out targets (253 participants), dynamic method achieved 60.865% equal-target balanced accuracy versus 52.684% for strict-clean 5/5 BIOT (hierarchical difference +8.182 points; 95% CI +6.382 to +10.196; positive in every target). On four EEGPT-clean targets, the descriptive means were 61.520% and 59.057%. In four post-hoc exact same-parent tests, FULL exceeded coordinate-basis, matched-spectrum, row-shuffled, and wrong-anatomy controls by 9.959, 9.698, 11.241, and 3.985 points, respectively (all 95% CIs excluded zero; Holm-adjusted P<0.001). A separate 60-participant same-paradigm evaluation retained a positive physical-operator contrast (+4.029 points; +2.338 to +5.773), whereas dynamic--static was inconclusive; its preregistered status is partial confirmation rather than unrestricted external validation. The SSM added +1.059 points (+0.013 to +1.892). Claims remain bounded to five binary motor-imagery domains and the tested operators.

0.14

Domain-specific language alignment of multimodal embeddings towards explainable predictions

cs.LG cs.CV ·2026-10-16 ·Gabrielė Tijūnaitytė, Thijs L. van der Plas, Mohamed-Bachir Belaid, Rob Knapen, Maciej J. Soja +1 more
Abstract

Multimodal data are commonly fused into latent representations that improve downstream task performance but are not interpretable, limiting their application in domains that require explainable model decisions. Vision-language alignment has made image embeddings interpretable through language, but requires vast image-caption training data. However, captioned data are typically not available for multimodal embeddings, as relevant features cannot directly be observed and annotated from embeddings alone. Here, we propose a framework to language-align any embedding space that can be linked to interpretable auxiliary data. Captioning functions convert auxiliary data into partial captions; a soft contrastive loss weights pairs by the similarity of their properties that each caption describes; and a rank-based concept retrieval task evaluates alignment against the auxiliary data rather than the captions. We apply this framework to multimodal Earth embeddings to facilitate explainable predictions for two species distribution modelling tasks: S2BMS and SatBird. AlphaEarth embeddings beat previous state-of-the-art models on 6/7 metrics, while our framework language-aligns domain-specific concepts of the embedding space, outperforming previous Earth observation vision-language models on average concept retrieval for both datasets. Finally, aligned concepts recover species-habitat associations ($ρ$=0.44) and relate habitat descriptions to species predictions, showing that abstract embeddings can be interrogated in the language of domain experts.

0.14

FuzzInsight: Navigating the Efficiency-Effectiveness Trade-off in Coverage-Guided DNN Fuzzing

cs.AI ·2026-10-16 ·Vivek Vekariya, Fathiyeh Faghih, Alexander Pretschner
benchmark
Abstract

Collecting and labeling diverse test datasets for evaluating Deep Neural Networks (DNNs) is costly and time-consuming. This has motivated coverage-guided fuzzing techniques that generate defect-revealing corner cases from limited test data. However, their effectiveness depends heavily on configuration choices, particularly coverage criteria and seed-selection strategies. We present a configuration-oriented analysis and decision guidelines for scalable DNN fuzzing targeting two bottlenecks: inefficient seed utilization and the cost of coverage guidance. Our analysis shows that existing seed-selection strategies repeatedly prioritize a small subset of inputs, leaving many collected seeds unexplored. To address this issue, we propose Equalized Sampling, a constraint-based strategy that increases mean seed queue utilization to 77%, improving on prior approaches by 13 percentage points. We further evaluate 84 fuzzing configurations on four benchmark datasets: MNIST, SVHN, GTSRB, and CIFAR-100. Results show that the non-structural criteria LSCD, LSC, and MDSC achieve coverage outcomes comparable to structural and alternative non-structural criteria while requiring substantially less computation. These criteria also promote broad behavioral exploration of the DNN under test, reducing the need to monitor multiple coverage metrics. Among them, LSCD provides a modest but consistent runtime advantage. Finally, we show that retraining with generated corner cases improves generalization robustness by up to 77 percentage points on average without degrading performance on the original test distribution. Together, these findings provide a practical blueprint for selecting fuzzing configurations that balance efficiency, behavioral exploration, and model hardening.

0.14

Flash RNN-T: Memory-Efficient Joint Network and Transducer Loss Computation

cs.LG cs.DCcs.SD ·2026-10-16 ·Mahmoud Ashraf
Abstract · Will be submitted to Interspeech 2027

The recurrent neural network transducer (RNN-T) remains a practical choice for streaming speech recognition, but training requires large intermediate tensors. We present Flash RNN-T, a joint and loss implementation combining specialized kernels with optional tiled recomputation over the full alignment space and vocabulary. Tiled and untiled Flash form the entire measured throughput-memory frontier on medium and long workloads. On H100, untiled Flash achieves 24-39% higher throughput with 26-56% less memory than the fastest baseline on these workloads wherever it fits. Tiling reduces memory further through recomputation. On medium, tiled Flash uses ninefold less memory than that baseline at 16% lower throughput when both use one backward call per batch. On long, it fits where all tested alternatives with that backward schedule run out of memory with 80 GB. These comparisons use each implementation's lowest-memory configuration reaching at least 95% of its peak throughput.

0.14

Budgeted Bayesian Bandits for Future Utilization

cs.LG cs.DS ·2026-10-16 ·Sudipto Guha
Abstract

The multi-armed bandit problem models the trade-off between exploration and exploitation. In this paper, we study the variant where exploration has a hard budget and precedes exploitation. Prior guarantees include a 1/3 approximation by Guha and Munagala using Lagrangian duality, and a 4/27 approximation by Ma against a primal linear programming relaxation for a more general model. We show that the analyses of both algorithms are tight, even when their relaxations are asymptotically exact. We provide a different efficient algorithm that achieves a 0.43 approximation. We then provide an efficient algorithm whose guarantee approaches 1-e^{-1} approx 0.63 for bounded exchangeable reward observations in a non-vanishing regime, which is asymptotically tight against natural linear programming relaxations. The technique also gives certificates for stopping sequential model evaluation.

0.14

Darwin-Claw: Environment-Adaptive Experience Consolidation for Recursively Self-Evolving Agents

cs.AI ·2026-10-16 ·Mingze Yin, Xiaohan Wang, Dian Li, Longfei Luo, Jiahong Yan +2 more
Abstract

While Large Language Model (LLM) agents have demonstrated remarkable capabilities across complex tasks, they predominantly operate in isolated sessions, lacking the ability to continually learn from historical interactions or adapt to user-specific environments. Existing self-evolution paradigms often indiscriminately inject raw interaction trajectories into memory or skill modules, leading to obscured behavioral patterns and compromised update controllability. Furthermore, as foundational LLM reasoning improves, the bottleneck for agentic evolution shifts from accumulating generic knowledge to effectively managing situational experience. To address this, we propose Darwin-Claw, an environment-adaptive self-evolving framework that systematically distills episodic experiences to drive the comprehensive evolution of operational skills and durable memory. Theoretically grounded in an iterative agent-actor and evolver-critic co-adaptation, Darwin-Claw introduces a Hierarchical Experience Consolidation System that extracts fine-grained error attributions and generalizes them into dynamically updated prototype vectors. This structured situational experience serves as interaction patches for context-sensitive perception and explicit optimization signals for skill refinement. Practically, Darwin-Claw is a readily deployable, automated framework compatible with mainstream agent systems without requiring manual curation. Extensive experiments on WildClawBench demonstrate that Darwin-Claw yields substantial performance improvements, establishing a controllable and stable pathway for the continual, user-centric adaptation of autonomous agents in real-world ecosystems.

0.14

Efficient CPU-GPU collaborative MoE inference via Adaptive Expert Assignment

cs.AI ·2026-10-16 ·Wenhui Lv, Yunming Liao, Hongli Xu, Yang Xu, Junhao Cheng
Abstract · 13 pages, 7 figures

Mixture-of-Experts (MoE) models reduce inference computation through sparse expert activation, but their large expert memory footprint often exceeds GPU capacity, making expert offloading necessary for memory-constrained inference. However, existing systems typically handle GPU-missed experts by either transferring them to the GPU, incurring substantial PCIe transfer latency, or executing them on the CPU, making the slower CPU computation the bottleneck. To address these limitations, we present \method, a CPU-GPU collaborative MoE inference system that adaptively assigns these missed experts to the CPU or GPU for efficient execution. By exploiting differences in expert token-loads, \method selectively transfers high-load experts to the GPU while executing low-load experts on the CPU. To minimize inference latency, \method adaptively adjusts expert assignments to balance the CPU computation path against the GPU transfer-and-computation path, allowing both execution paths to complete as early as possible. Extensive experiments on three representative MoE LLMs show that \method improves prefill speed by up to 51.9\% and decode throughput by up to 39.5\% over the strongest baseline across the evaluated batch sizes, with consistent gains under different GPU-memory budgets.

0.14

Dynamical Kernel Attention: Estimating the Present from a Record of the Past

cs.LG ·2026-10-16 ·Peter Racioppo
Abstract

Standard self-attention compares tokens by similarity but does not account for how much of a past state survives to the present. We introduce Dynamical Kernel Attention (DKA), which derives attention from robust estimation of a latent state evolving under stochastic dynamics. As a past state decays toward equilibrium under the dynamics, the information it carries about the present vanishes, bounding the total evidence a head can draw. Unlike recency biases, DKA can abstain when relevant evidence is weak and guarantees a bounded error when old tokens are discarded. DKA preserves the parallel structure and computational cost of standard attention. We present the model and its theory; empirical evaluation is deferred to a subsequent version.

0.13

Algorithm Modelization: Transforming Optimization Algorithms into Evolvable Intelligence

cs.AI cs.NE ·2026-10-16 ·Jian Yao, Zhiyao Zhang, Yichen Li, Xingyu Wu, Kay Chen Tan
Abstract

Can algorithms be learned and evolved as models, rather than repeatedly redesigned as programs? We explore this question by introducing Algorithm Modelization, a new paradigm that turns iterative algorithms into trainable models. Its central idea is to regard algorithm execution trajectories as computational experience: by learning how algorithmic states evolve over iterations, a model can internalize the knowledge embodied in existing algorithms, reproduce their computational behaviors, and further evolve beyond them through training. As a first realization of this paradigm, we modelize iterative optimization as an autoregressive process, where preceding generations constitute the computational context for generating the next population. We develop a dimension-independent population-temporal architecture to capture both population interactions and iterative dynamics, together with an imitation-to-surpassing learning framework that first acquires algorithmic knowledge from heterogeneous teachers and then improves the resulting optimizer directly through optimization feedback. The modelized optimizer consequently goes beyond reproducing individual algorithms: it consolidates heterogeneous search knowledge into a single trainable model, achieves stronger optimization capability through subsequent improvement, and generalizes its acquired knowledge across problems and substantially larger unseen dimensions. More fundamentally, Algorithm Modelization makes algorithmic capability accumulable through learning, where new algorithms and problem-solving experience can become training resources for progressively improving the model. Our results establish a promising path from manually designed algorithms toward general algorithm models that learn, generalize, and evolve from computational experience.

0.13

Morpheus-2.0: Multiscale Probabilistic Forecasting with Dual-Axis Attention

cs.LG cs.AIstat.ML ·2026-10-16 ·Aakriti, Manoj Cherukumalli, Prathamesh Patil, Leonidas Tsepenekas, Amit Varshney +3 more
Abstract · 26 pages, 5 figures, 16 tables. Aakriti and Manoj Cherukumalli contributed equally

A universal forecaster is trained once and applied everywhere, so it must serve tasks whose schemas, forecast horizons, and temporal resolutions all differ. The number of targets and covariates changes from dataset to dataset, horizons vary by orders of magnitude, and sampling rates range from seconds to years. Existing time-series foundation models address these only in part, since they isolate each series or visit the two grid relations in separate blocks, and they fix the horizon and resolution before training begins. We introduce Morpheus-2.0, a compact encoder-only foundation model that represents targets and covariates on a single variable-time grid. Its dual-axis attention runs both relations side by side in every layer, one across variables at a timestep and one across time within a variable; a learned gate weights the two per layer, and an axis-aware rotary encoding preserves both grid coordinates. The forecast length is sampled during training, so one set of weights spans short to long horizons at no additional inference cost. Morpheus-2.0 Multiscale extends the model to several temporal resolutions at once, weighting them according to the series being forecast. Across GIFT-Eval, which spans several domains and frequencies, the two models give complementary operating points: Morpheus-2.0 attains 0.685 relative MASE with 81M parameters, while the larger Morpheus-2.0 Multiscale attains 0.646 relative MASE and 0.455 relative CRPS, the lowest of any published system on both metrics.

0.13

GR-OPD: Group-Relative Sequence Credit for On-Policy Distillation

cs.LG ·2026-10-16 ·Yushu Wang, Renkai Yang, Jingcao Xu, Zili Geng, Haihong Tang +2 more
Abstract

On-policy distillation (OPD) provides token-level teacher feedback, yet common sampled-token updates assign credit locally and omit the effects of earlier actions on subsequent feedback. We introduce \method{} (Group-Relative On-Policy Distillation), which decouples feedback computation from credit assignment by aggregating token feedback into calibrated sequence-level credit. We establish a theoretical basis for this approach by showing that, under ideal on-policy assumptions, broadcasting raw summed feedback to every response token yields an unbiased estimate of the negative gradient of the sequence-level reverse-KL objective. To support effective learning in practice, \method{} discounts later feedback that may become less reliable under distribution shift and uses same-prompt rollout groups to calibrate teacher guidance. It preserves relative response preferences while retaining a controlled fraction of the group mean, which we term collective teacher pressure. Across five Qwen3 teacher--student settings and six mathematical reasoning benchmarks, \method{} outperforms vanilla OPD in macro Avg@K in every setting, including gains of 10.9 and 11.5 percentage points in two settings with matched thinking patterns. It also maintains stable training under mismatched thinking patterns, where vanilla OPD collapses. Ablations show that retaining an appropriate amount of collective teacher pressure can improve learning, while excessive collective teacher pressure or an unrestricted feedback horizon can destabilize training.

0.13

DeDiCa: A Scalable Deconfounding Discrete Causal Generative Model

cs.LG ·2026-10-16 ·Kunchangtai Liang, Adrián Javaloy, Antonio Vergari, Isabel Valera
Abstract · 32 pages, 14 figures, 7 tables

We introduce DeDiCa, a practical framework for efficient causal inference with discrete variables and hidden confounding that revisits classical probabilistic modeling through a tensorized parameterization of causal Bayesian networks, enabling tractable inference. Given a causal graph, DeDiCa is trained once on observational data by maximum likelihood, exactly marginalizing over hidden confounders, and can then estimate any (proxy-)identifiable interventional query. Despite this deliberately simple parameterization, DeDiCa matches state-of-the-art neural causal generative models in ATE estimation, while supporting memory-efficient CPU training and exact interventional inference in milliseconds. Compared with causal foundation models, DeDiCa scales more effectively with additional observational samples and achieves competitive accuracy at substantially lower inference time and memory. DeDiCa provides a tractable, accurate, and scalable approach to causal inference in discrete settings with hidden confounding.

0.13

Data-Efficient Classifier Pretraining Through Diffusion Models

cs.LG ·2026-10-16 ·Qinyu Li, Yee Whye Teh, Chris Holmes, Joel Jaskari
Abstract

Machine learning with limited data is susceptible to overfitting, especially with modern deep neural networks. Prior to the introduction of AlexNet, reconstruction-based pretraining with auto-encoder and denoising auto-encoder were found to be beneficial, partially due to the limited size of the training sets at the time. Motivated by these traditional approaches, we propose DiffPT, a simple diffusion modeling approach to pretrain a given classifier backbone for limited domain-specific training data. We experiment on time-series data, standard natural-image benchmarks, and medical imaging datasets, and contrast the diffusion approach to a multitude of modern and traditional un/self-supervised methods. While contrastive pretraining performs better on CIFAR-10 and Tiny ImageNet in low-data regimes, DiffPT is particularly effective on medical imaging and time-series tasks, where it outperforms self-supervised baselines across a range of low-data regimes. Our results suggest that DiffPT is a simple and effective approach for data-efficient classifier pretraining on different modalities.

0.12

Sep-Latent: Separator Tokens as Natural Compression Anchors for Implicit Reasoning

cs.AI cs.CL ·2026-10-16 ·Yinlong Xu, Jianfeng Huang, Zhengdong Zhou, Mingjie Lin, Fuyu Lv +3 more
benchmark
Abstract

Large Language Models (LLMs) have demonstrated remarkable success in complex reasoning through explicit Chain-of-Thought (CoT), but the associated computational overhead and token verbosity remain significant barriers to efficiency. While implicit reasoning in continuous latent spaces offers a promising alternative, existing methods like Coconut and CODI often rely on fixed-length latent sequences and suffer from representational collapse. In this paper, we propose Sep-Latent, an adaptive implicit reasoning framework that utilizes separator tokens as structural semantic anchors. By introducing an Attention-Decay mechanism, we force the model to aggregate and compress reasoning logic into these anchors during training. We then align explicit and implicit pathways through multi-level distillation across all layers to ensure logical consistency. Crucially, Sep-Latent enables dynamic reasoning depth by adaptively terminating the latent thinking phase based on query complexity. Experimental results on the GSM8k-Aug benchmark show that Sep-Latent achieves an accuracy of 56.8\% with an average of only 5.4 tokens, significantly outperforming prior fixed-length implicit models in efficiency while narrowing the performance gap with explicit CoT. Further analysis via PCA and symbolic variant testing confirms that our method ensures semantic grounding and exhibits superior robustness to perturbations.

0.12

Dynamic Neural Optimal Transport for Subset Alignment

cs.LG ·2026-10-16 ·Austin J. Brockmeier, Bilal Riaz
Abstract

We study optimal transport between distributions when only a subset of the target population is relevant to the source. The goal is to map the source distribution to a reweighted target distribution whose density ratio with respect to the full target lies in $[0, c]$. This constraint enables the transported distribution to select an appropriate target subpopulation while limiting how strongly any region of the target can be emphasized. We propose a dynamic neural optimal transport framework for this setting using two neural networks. The first parameterizes a time-dependent potential whose spatial gradient defines the velocity field and whose terminal value enforces the density-ratio constraint. The second parameterizes an interpolation between the source distribution and the learned terminal distribution. Unlike prior neural dynamic optimal transport methods, which typically assume a fixed terminal distribution, our formulation learns the terminal distribution within the support of the target. We demonstrate the method on positive-unlabeled learning, unpaired domain translation, and trajectory inference, where meaningful alignments exist between source and target distributions, but direct distribution matching is inappropriate because the target contains additional or irrelevant components

0.12

Evo-B&B: Behavior-Aware LLM-Guided Joint Evolution of Branching and Node Selection Policies for MILP

cs.AI ·2026-10-16 ·Jiayu Chen, Chentong Chen, Yuhan Liu, Ye Fan, Jianyong Sun +1 more
Abstract

In branch-and-bound for Mixed-Integer Linear Programming (MILP), branching and node selection are two critical and tightly coupled decisions, yet most existing learning-based approaches focus on optimizing them separately. We propose Evo-B&B, an LLM-guided evolutionary framework that jointly evolves executable branching and node selection policies using end-to-end SCIP solving feedback. To improve search efficiency and reduce redundant evaluations of behaviorally similar policies, Evo-B&B introduces a behavior-aware multi-island evolutionary mechanism. It measures policy similarity from branching and node-selection behaviors on a fixed set of B&B decision states and uses this similarity to preserve population diversity. We evaluate Evo-B&B on four standard MILP benchmarks across multiple instance scales. Experimental results show that Evo-B&B achieves competitive solving performance and generalizes well to larger-scale instances, while ablation studies further demonstrate the effectiveness of joint evolution and behavior-aware diversity management.

0.12

Hidden Trajectories Don't Lie: Correcting Corruption Drift for Test-Time Adaptation

cs.LG ·2026-10-16 ·Amirhossein Sabour, Amir Khasahmadi, Mengliu Zhao
Abstract · 35 pages, 15 figures

Test-time adaptation (TTA) adapts pretrained models to distribution shifts encountered during deployment. However, most existing methods rely on the weight representations of the final layer or signals derived from the model's output probability distribution. We study hidden trajectories: the ordered sequence of intermediate representations produced across a deep neural network. We observe that clean and corrupted trajectories become more separable as corruption severity increases. We introduce Drift-TTA, which adds zero-initialized low-rank adapters to a frozen backbone and trains them to align corrupted hidden trajectories with those from cleaner inputs. The approach supports two variants: a source-dependent version trained with clean targets, and a source-free setting that learns from corrupted images at different severities. We develop an online TTA approach that freezes the backbone and adapters while updating a scalar gate to control adaptation strength. We further cast trajectory alignment as MAP inference under a Gaussian observation model and derive a KL-based transfer bound characterizing test-time trajectory misalignment in terms of offline residual error, residual dimensionality, and distribution shift between offline and test corruptions. We evaluate Drift-TTA on CIFAR-10-C, CIFAR-100-C, and ImageNet-C using supervised ViTs and CLIP models. On CIFAR-100-C at severity 5, our source-free online approach improves accuracy over no TTA by 13.3 percentage points with ViT-B/16 and 11.6 percentage points with CLIP ViT-B/16, reaching 74.9% and 47.4%, respectively. Under the same setting, prior TTA baselines yield ECEs of 18.00% for ViT and 14.70% for CLIP, while our approach reduces them to 9.53% and 5.77%, respectively. These results demonstrate that hidden trajectory alignment provides an effective signal for both offline correction and lightweight online test-time adaptation.

0.12

Continual Test-Time Learning with Evolution Strategies

cs.LG cs.CL ·2026-10-16 ·Guangya Hao, Yiting Duan, Longyan Tan, Yunbo Long, Hao Wu +3 more
Abstract

Large language models increasingly operate in open-ended environments, where they are expected to continually adapt to evolving tasks and distributions, often without access to external supervision. Existing continual learning approaches rely on external supervision, such as labeled data, demonstrations, or reward signals, which can be costly or unavailable in practice. Test-time adaptation removes this dependence by learning from self-generated supervision, but existing methods derive that supervision from the single policy being updated, so once the policy drifts, its own learning signal drifts with it, leading to policy collapse, degraded reasoning, and catastrophic forgetting. In this paper, we propose \textbf{Evolution Strategy Reinforcement Learning (ESRL)}, a continual test-time learning framework that draws its self-generated supervision from a population of policies instead of the single policy being trained. ESRL perturbs an anchor policy in parameter space to form a population of nearby policies, samples one trajectory from each, and aggregates their answers into label-free group-relative learning signals for the central student. The anchor is held fixed within each block and refreshed from the student at block boundaries, so the population stays centered on a stable yet current policy and continues to supply diverse and reliable supervision as the student moves. In single-task test-time adaptation across mathematical reasoning, general knowledge, and medical domains, ESRL improves average accuracy by up to 31.7% relative to the base model and 16.6% relative to the strongest test-time adaptation baseline. Notably, ESRL substantially reduces catastrophic forgetting and achieves the strongest final performance in continual learning.

0.11

World Action Critic (WAC) Model: Joint Foresight and Criticism for Policy Learning

cs.AI ·2026-10-16 ·Xiangteng Zhang, Haoyuan Liu, Yang Guan, Jingliang Duan, Yao Lyu +5 more
Abstract

World models have shown considerable promise in sequential decision-making by anticipating the consequences of actions and providing predictive foresight for policy learning. However, learning to reason about counterfactual actions remains challenging, since predictive objectives may underemphasize information critical for action evaluation, while unexecuted actions lack observed futures for direct supervision. To address these challenges, we propose the World Action Critic (WAC) Model, which proposes multimodal action candidates and jointly learns candidate-conditioned future prediction and consequence evaluation. In our proposed framework, critic supervision directly shapes intermediate predictive representations during rollout, encouraging them to retain information needed to distinguish action quality. Observed futures supervise prediction for executed actions, while external task-quality targets provide evaluation signals for counterfactual candidates. Through joint optimization, critic signals from these candidates propagate into predictive representations without requiring their observed futures. Our theoretical analysis characterizes critic-guided feature retention and bounds candidate-selection regret. Applied to autonomous driving, WAC achieves 93.4 PDMS on NAVSIM-v1, demonstrating the effectiveness of jointly learning predictive foresight and consequence evaluation.

0.11

Out of Support, Not Out of Reach: In-Context Learning with Unsupported Instructions and Composed Tasks

cs.LG ·2026-10-16 ·Xuhan Tong, Yuchen Zeng, Jiawei Zhang
Abstract

In-Context Learning (ICL) enables pretrained LLMs to adapt to downstream tasks from a small set of input-output demonstrations, without parameter updates. Yet existing theory primarily explains ICL as identifying a latent task already represented in pretraining, leaving unclear whether ICL can succeed when either the instruction or the target task falls outside the pretraining support. We develop a theoretical framework under mild assumptions that separates these two forms of distribution shift. For target tasks represented in pretraining, we show that demonstrations can identify the task and reduce sensitivity to pretrain-supported prompt templates as their number increases. For an unsupported instruction, we show that under a fixed model-induced reference extension, demonstration evidence can dominate an $N$-independent instruction shift while preserving the exponential identification rate. We then use the same ICL framework as an interpretive lens for long-horizon target tasks that are not represented in pretraining but admit a plan that decomposes them into pretraining-supported subtasks. Applying the supported-task analysis step by step yields a standard accumulated-error bound and clarifies how invoking familiar subtasks along an intermediate-state trajectory can avoid directly selecting the unsupported complete task. Together, our results characterize when demonstrations, instructions, and task decomposition can support ICL beyond the corresponding pretraining supports.

0.11

A Stronger Teacher Within: Strengthening MoE Models via On-Policy Bidirectional Distillation

cs.LG cs.AI ·2026-10-16 ·Haoyu Chen, Jun Bai, Yang Liu, Jianfei Zhang, Wenge Rong
Abstract

Reinforcement Learning with Verifiable Rewards (RLVR) has substantially improved Mixture-of-Experts (MoE) LLMs, but its outcome-level rewards provide only sparse supervision. On-Policy Distillation offers dense token-level feedback, yet requires a stronger compatible teacher, while on-policy self-distillation relies on hand-designed privileged information. We observe that the MoE architecture itself offers a way to resolve this dilemma, as increasing the activation budget per token exposes latent expert capacity and yields a potentially stronger teacher within the model. However, naive expansion degrades performance because newly co-activated experts are not optimized to work together. We therefore propose MoEvolve, a bidirectional distillation framework following a recovery-then-improvement strategy: we first restore the activation-expanded high-$k$ model and further strengthen it by exploiting its additional active capacity, then reverse the direction of distillation to transfer the enhanced capabilities back into the original low-$k$ configuration, retaining its inference-time activation budget. Across diverse MoE models and benchmarks, MoEvolve improves the initial models by up to $22.2$ domain-average points and outperforms same-budget RLVR baselines by up to $6.5$ points. Notably, the final low-$k$ models often surpass their activation-expanded teachers.

0.11

Rethinking GRPO for Large Reasoning Model Safety Alignment

cs.AI ·2026-10-16 ·Jiawei Chen, Yi Zhang, Jiawei Du, Zhonghou Lv, Tairan Chen +7 more
Abstract

Large reasoning models (LRMs) exhibit strong reasoning capabilities but remain vulnerable to unsafe outputs. While Group Relative Policy Optimization (GRPO) is widely used for safety alignment, its group-relative optimization may provide insufficient safety-directed learning signals. In this paper, we identify two failure modes: advantage collapse, where homogeneous rollout groups yield vanishing advantages, and unsafe diversification, where unsafe behaviors persist and become increasingly diverse. Our theoretical analysis reveals that behavioral consistency weakens within-group safety contrasts, while entropy regularization can diversify unsafe responses when relative signals vanish. To address these challenges, we propose Beacon, comprising Safety Signal Recovery (SSR) and Safety-Utility Decoupled Alignment (SUDA). SSR combines advantage-guided trajectory filtering and safe reference supervision to recover effective safety learning signals, while SUDA employs category-aware rewards and benign-only KL anchoring to mitigate over-refusal without compromising safety. Experiments across four Qwen3 and DeepSeek-R1-Distill variants and eight benchmarks show that our framework achieves the best average safety and reasoning performance, reducing average safety error by up to 32.83 percentage points while maintaining low over-refusal.

0.11

CoCoNet: Convex Conformal Network for Multi-Dimensional Uncertainty Regions

cs.LG stat.ML ·2026-10-16 ·Alireza Moradi, Thomas Bruys, Pascal Van Hentenryck, Reza Zandehshahvar
Abstract

Conformal prediction (CP) provides distribution-free coverage guarantees under exchangeability, but the efficiency of its prediction regions depends critically on the choice of the conformity score. This is particularly important in multivariate regression, where dependencies across output dimensions induce complex joint uncertainty structures. This paper introduces CoCoNet (Convex Conformal Network) for learning convex conformity scores that induce flexible, data-adaptive prediction regions. CoCoNet learns the score from held-out data by optimizing prediction-region efficiency while enforcing convexity with respect to the residual. CoCoNet then applies split-conformal calibration to retain finite-sample coverage guarantees. The paper presents two variants of the method: CoCoNet$_{\mathrm{Spline}}$, based on piecewise-linear scores, and CoCoNet$_{\mathrm{ICNN}}$, based on Input Convex Neural Networks. Experiments compare CoCoNet with other CP baselines over synthetic and real-world datasets with output dimensions ranging from 2 to 16. The results demonstrate the flexibility of CoCoNet in learning different response types. Across 8 models and 30 multi-dimensional regression tasks, CoCoNet$_\text{ICNN}$ and CoCoNet$_{\mathrm{Spline}\text{-}\ell_1}$ achieve the best efficiencies with average ranks of 1.45 and 2.75, respectively. In particular, CoCoNet$_\text{ICNN}$ ranks first on 22 tasks out of 30 in terms of efficiency.

0.11

FuTransformer: FP8 prefill and INT4 decode for Llama-style inference on Hopper

cs.AI ·2026-10-16 ·Wei Wang
Abstract

Transformer prefill reuses weights across tokens and is dominated by matrix multiplication and activation data movement; single-token decode has limited reuse and is dominated by weight and KV-cache reads. We present FUTRANSFORMER, a CUDA implementation of a Llama-style Transformer block for NVIDIA Hopper GPUs. It applies CODA's GEMM-plus-epilogue approach to inference, adding direct KV-cache placement, folding the fixed RMSNorm gain into weights, writing FP8 output directly with delayed scaling, and supporting prefill without padding. Decode uses INT4 projection weights and an INT4 KV-cache copy. Weights and keys follow Tensor Core fragment layouts for direct dequantization into MMA operands without full intermediate tiles. A block-diagonal Hadamard transform reduces cache quantization error from outlier channels. On one H100, prefill is up to 1.48x faster than compiled PyTorch FP8 and reaches 1.29 PFLOP/s. For gate/up projection with SwiGLU at the 8B shape, our BF16 path is 1.13x faster than CODA. Our FP8 path is 2.02-2.18x faster than the evaluated BF16 CODA path. Across 8B and 70B shapes at batches 1-32, INT4 decode is 1.3-2.4x faster than the faster of PyTorch's BF16 and INT4 blocks at each point: 2.1-3.3x faster than compiled, CUDA-graphed BF16 and 1.3-6.3x faster than the same block using PyTorch's INT4 projection kernel, which targets a single query row and loses to BF16 at batch 32. It is also 31-51% faster than our FP8 path. Complete Llama-3.1-8B decode takes 2.8 ms per step at batch one, versus 8.1 ms for the compiled transformers implementation. Decoding every WikiText-2 token through the INT4 weight path raises perplexity from 6.24 to 7.23 with round-to-nearest weights or 6.99 with GPTQ in the same format; the rotated INT4 cache adds 0.20. The kernels match a quantized reference computation within 0.001 perplexity.

0.11

Who Speaks, Who Authorizes, Who Acts? Understanding Trust Boundaries in Multi-Agent LLMs

cs.AI ·2026-10-16 ·Junru Zhu, Xin Chen, Hang Lyu, Jiayue Meng, Muzi Li +4 more
Abstract · 14 pages, 4 figures, 9 tables

Multi-agent LLM systems exchange messages before acting through tools, but message delivery, signing authority, and tool-side enforcement are separate trust boundaries. We represent them as communication, authorization, and execution structures, and use the authority cut to characterize the smallest independently compromised signer coalition that can construct a policy-valid false-claim proof under stated assumptions. Controlled paired hidden-state studies reveal four distinctions. First, no fixed-message routing contrast across four models meets the prespecified large-effect criterion; when relays rewrite messages, executed false-state action rates vary by model and route configuration. Second, presented source-identity cues do not uniformly increase protected-action selection across three models. Third, on invalid-proof cases, Nemotron and Llama request protected actions in 52.1% and 54.2% of trials, respectively, yet a deterministic tool gate commits none: request-level and commitment-level safety readouts diverge. Finally, a separate single-model authorization check shows that a false claim with a policy-valid compromised-signer proof can pass the gate, since proof validity is not factual truth. These findings motivate evaluating delivered information, qualified authorization, model proposals, and committed actions separately.

0.11

Where Does the Performance Go? Understanding and Recovering Performance Regressions in C-to-Rust Translation

cs.AI cs.PL ·2026-10-16 ·Zhiqiang Yuan, Chong Wang, Zhenhao Zhou, Junwei Liu, Xin Peng
Abstract

Rust is increasingly adopted as a memory-safe successor to C, and transpilers such as \crust{} can migrate entire C projects while preserving their behavior. Preserving behavior, however, does not guarantee preserving performance. We study this gap on 20 real-world C projects and their \crust{} translations. On the eight projects used for analysis, the translated Rust is more than 3\% slower than the original C on 12 of 19 operations, by up to 62.2\%. By comparing execution-cost profiles, optimized LLVM IR, and compiler optimization remarks across each C--Rust pair, we trace these slowdowns to nine recurring performance issues, spanning residual runtime work, missed compiler optimizations, and C-style source patterns, and derive an optimization direction for each. We then present \ourtool{}, an LLM-based approach that encodes these issues as reusable optimization rules to recover the performance of translated Rust. \ourtool{} profiles each project to locate hot functions and matches them against rule-specific patterns. The LLM then checks each rule's rewrite preconditions against the project context before rewriting, and a rewrite is kept only if it passes both functional and performance validation. On 41 operations from twelve held-out projects, \ourtool{} preserves C outputs on all tested workloads and speeds up 40 operations over the \crust{} translations, 18 of which end up faster than the original C. It outperforms the general-purpose coding agent \codex{} on 31 of the 37 operations where both preserve C outputs, and outperforms its own variant without rule guidance on 34 of the 41 operations.

0.11

Untrained CNN and coefficient-space regularization for joint reconstruction in photoacoustic tomography

cs.LG math.OC ·2026-10-16 ·Alfio Bozi, Anwesa Dey, Souvik Roy
Abstract

The joint reconstruction of the initial pressure and spatially varying sound speed in photoacoustic tomography from boundary measurements is considered. A regularized data-driven framework is proposed in which the unknown fields are represented in finite-dimensional wavelet spaces and fixed nonlinear features generated by an untrained convolutional neural network (CNN) are used to reconstruct the initial-pressure coefficients through Tikhonov-regularized least squares. A learned coefficient-space prior is subsequently used to infer the squared sound-speed coefficients. Numerical experiments demonstrate accurate reconstruction on held-out data and strong generalization to previously unseen phantom geometries. The frozen reconstruction map also remains stable under measurement noise not represented during training, with only minor changes in reconstruction accuracy for noise levels up to $10\%$. In comparison with the supervised trained-CNN approach considered in this work, the untrained representation yields substantially smaller reconstruction errors while requiring approximately one-sixth of the computational time. These results demonstrate that fixed random convolutional features combined with regularized coefficient-space learning provide an efficient and robust approach to joint reconstruction in photoacoustic tomography.

0.11

AgentOnto: Ontology-Guided LLM Agents for Enterprise Data Analytics

cs.AI cs.DB ·2026-10-16 ·Dayan Pan, Bincheng Zuo, Xiaojian Wang, Daogang Zhang, Jingyuan Wang
Abstract · Under review

Enterprise data analysis requires agents to connect analytical requests with knowledge distributed across databases, business documents, and existing example tasks. Without an explicit representation of this knowledge, an agent may produce incorrect analytical results and cannot reliably incorporate feedback. We present AgentOnto, an ontology-guided LLM agent for enterprise data analytics. AgentOnto organizes heterogeneous enterprise knowledge into an ontology that directly guides analysis and incorporates confirmed feedback. Iterative Ontology Preparation further improves the ontology through self-generated analysis tasks. We evaluate AgentOnto on two public data sources. AgentOnto substantially outperforms representative methods, while module-level evaluations confirm the effectiveness of ontology construction, analysis, and updating. A real-world deployment further demonstrates its practical value in enterprise analytics. These results demonstrate that explicitly constructing, using, and updating enterprise knowledge supports reliable and adaptable data analysis.

0.09

Predicting CVSS and CWE from CVE Descriptions with Fine-Tuned Small Language Models

cs.LG cs.CR ·2026-10-16 ·Tara Dixit, Jessie Dong, Summer Li
Abstract · 23 pages, 5 figures, 21 tables

Until April 2026, the National Vulnerability Database labeled nearly every published Common Vulnerabilities and Exposures (CVE) entry with a Common Vulnerability Scoring System (CVSS) v3.1 vector, a severity tier, and a Common Weakness Enumeration (CWE) weakness type. Now, NIST enriches only the high priority CVEs, leaving many newly published vulnerabilities unlabelled with the classifiers that security teams use for patch triage. In CVE enrichment, we predict the eight base components of CVSS v3.1 vector, base score, severity tier, and CWE from the vulnerability's textual description alone. To prevent leakage between near-duplicate descriptions, all systems are trained only on CVEs published in 2022 and 2023. Results are evaluated on all 36,325 usable 2024 CVEs in the pinned NVD snapshot. All scores are either directly predicted, or calculated from predicted vectors using the CVSS formula. A fine-tuned 3.8B-parameter model Phi-3.5-mini, trained with LoRA in 2.5 hours on a single Apple M5 Pro chip, achieves 69.9% accuracy on severity and 70.9% on CWE. It matches the official vector exactly for 48.8% of test CVEs, outperforming TF-IDF logistic regression by 2.8 to 4.7 points on all accuracy metrics based on zero-exclusion paired bootstrap intervals. When we evaluate the base model with a zero-shot prompt, it achieves 20.8% accuracy on severity in a sample of 2,000 CVEs. In this same sample, its stated scores match its own vector precisely for only 3.4% of CVEs. However, deriving scores from predicted vectors boosts the critical recall of TF-IDF logistic regression from 50.7% to 73.3% with no additional cost. And since the errors of the fine-tuned model and TF-IDF affect different CVEs, at least one system matches the correct severity tier for 75.8% of test CVEs.

0.09

Accelerating Wildfire Hazard Mapping Across Canada with an AI Surrogate Modelling Framework

cs.AI ·2026-10-16 ·Hager Radi Abdelwahed, Toluwaleke Olutayo, Chirag Goel, Jana Osea, Aldo Zaimi +5 more
Abstract · Preprint - under review

Canada's increasing wildfire activity requires frequent national hazard mapping to guide proactive mitigation, including fuel management, land-use planning, infrastructure protection, and evacuation routing. However, process-based wildfire simulation frameworks (e.g., BurnP3 and BurnP3+) rely on computationally intensive Monte Carlo simulations to model burn probability, fire intensity, and rate of spread. These workflows can take days to weeks to process a single regional modelling unit. To overcome these computational bottlenecks, we present three contributions developed through collaboration between AI researchers and wildfire modelling experts: (a) a standardized framework for training, fine-tuning, and evaluating AI surrogate models against process-based wildfire simulators; (b) a multi-task deep learning surrogate trained on national-scale data to emulate burn probability, mean fire intensity, and mean rate of spread at a fraction of the computational cost (seconds); and (c) domain-specific metrics and counterfactual evaluation protocols to probe out-of-distribution generalization. Integrating geospatial data (fuel, terrain, ignition likelihood) and regional weather, our model approximates BurnP3+ simulation outputs, achieving strong performance across agreement, rank-correlation, error, and spatial-overlap metrics. It preserves downstream hazard classifications and produces physically plausible responses to expert-informed counterfactual interventions. Furthermore, we demonstrate zero-shot transfer to a separate regional BurnP3+ study and show that lightweight fine-tuning can substantially improve regional calibration and predictive agreement. On a representative region, the surrogate achieves 806-fold end-to-end CPU acceleration and uses 13% of BurnP3+'s host memory, facilitating rapid scenario analysis.

0.08

Causal Discovery beyond Conditional Independence: Identifiability, Algorithms, and Statistical Inference for Directed Acyclic and Cyclic Graphs

cs.LG stat.MEstat.ML ·2026-10-16 ·Yang Ni
Abstract

Learning causal structure from data requires assumptions about mechanisms, noises, observation schemes, and interventions. Historically, causal discovery has relied primarily on conditional independence. This review focuses on modern causal discovery methods that exploit properties of distributions beyond conditional independence. Our organizing principle separates the source of identification from the statistical estimator and its computational implementation. This perspective connects conditional-independence methods with distributional asymmetries, conditional-moment restrictions, and variation across environments or individuals. We examine discovery under latent confounding and feedback, distinguishing causal information about observed variables from reconstruction of latent structure and emphasizing the additional semantics required by cyclic models. We also distinguish population identification, frequentist and Bayesian consistency, uncertainty quantification, and algorithmic guarantees. Throughout, examples show why the same observational distribution can support different conclusions under different model classes. We conclude with guidance for comparing methods at their identifiable targets and a research agenda for uncertainty, heterogeneous populations, non-scalar variables, and the intersection of discrete data, feedback, and hidden confounding.

0.07

Scalable long-range machine learning interatomic potentials with electrical response for atomistic material modeling

cs.LG cond-mat.mtrl-sci ·2026-10-16 ·Yicheng Chen, Xiaoyun Yu, Hyejeong Song, Xintian Wang, Zihui Li +3 more
Abstract

Accurate atomistic simulations of materials under electric fields require long-range electrostatics and electrical-response modeling at length scales beyond the reach of conventional first-principles methods. Yet these capabilities remain difficult to reconcile with the scalability of machine-learning interatomic potentials (MLIPs). Strictly local models enable efficient spatial parallelization but neglect long-range interactions, whereas existing electrostatic extensions introduce global dependencies that hinder distributed computation. Here, we overcome this limitation by analytically separating the long-range electrostatic forces and Born effective charges (BECs) into a global electrostatics calculation and strictly local derivatives of learned latent charges. This formulation preserves the spatial decomposability of equivariant MLIPs while enabling distributed evaluation of energies, forces, stresses, and BECs across different boundary conditions. Implemented as DPA4C-LR on the \texttt{nvalchemi} platform, the framework scales to 5.2 million atoms on four GPUs at approximately twice the computational cost of its short-range backbone. Trained on energies and forces, DPA4C-LR accurately infers BECs and reproduces the field-dependent infrared spectra of water. We further demonstrate its capabilities by recovering the characteristic slim polarization hysteresis of relaxor \ce{Pb(Mg_{1/3}Nb_{2/3})O3}, which is obscured by finite-size effects in smaller simulations; predicting a room-temperature field-driven ionic conductivity of $6.8 \pm 0.7$~mS/cm in amorphous \ce{Li_{0.9}NbO_{0.9}Cl_{4.1}}; and modeling interfacial thermal transport across Si/diamond interfaces with slab-corrected electrostatics. By combining long-range physics, electrical-response inference, and distributed computation, this framework extends predictive atomistic modeling toward device-relevant length scales.

0.07

Language, Vision & Media 235

Identify, Classify, Act: Per-Instruction Controlled Compliance for LLMs over External Data

cs.CL cs.CR ·2026-10-16 ·Zesheng Shi, Xue Li, Chenrui Dai, Jing Li
top venuepositioncode/data
Abstract

LLMs over external data face a per-instruction tension: a single document or tool return often interleaves legitimate directives with adversarial prompt injections, and the model must follow the former while rejecting the latter. Existing defenses suppress or trust all external instructions, collapsing the safety-utility frontier and forcing a choice between over-refusal and over-compliance. We argue the missing capability is \textbf{per-instruction credit attribution} and formalize it as External-Data Instruction Classification, an identify-classify-act decision over each embedded directive. We propose \textbf{EDIC}, a decomposed reward framework for per-instruction compliance. A scene-grounded pipeline injects mixed beneficial and harmful instructions with counterfactual filtering, while a generative judge is initialized by SFT and refined with GRPO to attribute compliance at the instruction level. The judge then produces a reward that separately encourages beneficial compliance and penalizes harmful compliance for RL training. Across two backbones and five benchmarks, EDIC improves both safety and utility over the strongest prior defense, lifting harmful-rejection by up to 7.7\% and Arena-Hard-Auto utility by 12.5\% on Qwen3-4B. The code repository is hosted at https://github.com/yingyingxia666/EDIC.

0.82

Cross-Centaur: Modeling Individual Decision-Making Across Psychological Tasks with Language Models

cs.CL ·2026-10-16 ·Hikaru Asano, Hidezo Suganuma, Ryutaro Mori, Yukino Baba
top venuedatasetcode/data
Abstract · Accepted to EMNLP 2026 (Main Conference, Oral Presentation)

Large language models (LLMs) are increasingly used as computational models of human behavior. Recent models such as Centaur have advanced this goal by fine-tuning LLMs on trial-by-trial decisions from psychological experiments. However, existing large-scale behavioral datasets do not link the same participants across different tasks, so these models learn population-level regularities but are not trained to model individual-level regularities across tasks. We introduce Cross-Psych-101, a dataset of 591 participants and 261{,}551 valid decisions across 12 Bandit, Tradeoff, Structure, and Social tasks. We train Cross-Centaur to predict a participant's target-task actions from a target-task description and the source-task description paired with the same participant's source-task behavior. Cross-Centaur outperforms off-the-shelf LLMs and Centaur on held-out participants, both within and across tasks. Probing analyses suggest that its hidden states encode participant-level behavioral tendencies that transfer across tasks. These results highlight the potential of cross-task individual-level data for predicting human decision-making with language models. Cross-Pysch-101 is available at https://huggingface.co/datasets/HikaruAsano/cross-psych-101.

0.64

MTEdit: Packed Trajectories Beat Rollouts for Multi-Turn Image Editing

cs.CV ·2026-10-16 ·Houyuan Chen, Fanzhou Wang, Xiangyu Fan, Songlin Yang, Zhitao Yang +6 more
benchmarkdatasetcode/datacommunity pick
Abstract · 41 pages, 22 figures, 8 tables

Multi-turn image editing aims to modify an image according to a sequence of instructions, making it better aligned with human-AI collaborative creation than single-turn editing. However, constructing high-quality multi-turn training data remains challenging, as conventional sequential rollout accumulates errors and introduces visual-quality drift across editing turns. To address this challenge, we introduce MTEdit, a multi-turn image editing dataset built through a pipeline centered on packed trajectory canvas (PTC) generation. By jointly generating multiple editing states within a unified visual context, this pipeline reduces visual-quality drift across turns. Covering both step-wise and context-aware editing scenarios, MTEdit comprises 500K multi-turn image editing trajectories, with up to 8 turns and an average of 5.38 turns per trajectory, making it one of the most extensive datasets for multi-turn image editing. We further introduce MTEdit-Bench, a benchmark designed to evaluate both scenarios by assessing models' ability to follow editing instructions, maintain content consistency, resist visual-quality drift, and effectively use historical context across turns. Using MTEdit, we fine-tune Qwen-Image-Edit-Causal to obtain Qwen-Image-MTEdit. On MTEdit-Bench, we evaluate a broad range of closed-source and open-source image editing models, including Qwen-Image-MTEdit. Experimental results show that Qwen-Image-MTEdit achieves the best performance among the evaluated open-source models, demonstrating the effectiveness of MTEdit in improving models' multi-turn image editing capabilities.

0.59

Language Models Learn to Improve Language Models

cs.CL cs.AI ·2026-10-16 ·Minhua Lin, Zijun Wang, Zewen Liu, Zhaotian Weng, Bing He +5 more
Abstract · 29 pages, 12 figures, 14 tables

Large language models (LLMs) are increasingly used to automate AI research, including language-model (LM) development, offering a potential path toward recursive self-improvement. However, turning experimental feedback into a learned ability to propose effective improvements across models faces two challenges. First, training such an ability with real feedback is expensive: reinforcement learning (RL) requires evaluating tens of thousands of proposals, each potentially requiring a full model-training run. Even at ten GPU-hours per run, reward collection alone would consume hundreds of thousands of GPU-hours, motivating a cheaper surrogate. Second, surrogate feedback can become unreliable as policy optimization favors overestimated proposals when the proposal distribution shifts. To address these challenges, we propose LM2LM, which trains LLMs with RL to generate effective LM improvement proposals. To provide scalable feedback, we use a research world model (RWM), a frozen LLM that predicts proposal gains from real experimental records provided in context, as the reward model for RL. To maintain feedback reliability, we periodically execute current-policy proposals ranked highly by the RWM and add their measured outcomes to its context. Across eight autoregressive and diffusion LM development environments, real-execution evaluations on held-out tasks show that LM2LM enables Qwen3.8-27B to outperform GPT-6-sol, even when GPT-6-sol is given 300 real experimental records in context, while requiring 33% lower data-generation cost. LM2LM also generalizes to a model family unseen during training.

0.56

Learning to Correct Geometry in Generated Videos

cs.CV ·2026-10-16 ·Weijie Lyu, Tiancheng Shen, Jason Li, Yujing Wang, Xiaoyuan Wang +3 more
top venuepositioncode/data
Abstract · Accepted by NeurIPS 2026. Project page: https://weijielyu.github.io/VGC

Generated videos often lose geometric consistency under camera motion: structures warp under viewpoint change, parallax becomes inconsistent, and smooth camera movement may collapse into shot-like transitions. We argue that this failure is largely a data problem. In web-scale video corpora, most visible motion comes from dynamic objects captured by static or weakly moving cameras, leaving only sparse supervision for viewpoint-induced appearance changes. To address this, we propose Video Geometry Corrector (VGC), a post-hoc video-to-video model trained on synthetic paired supervision. Starting from real videos with diverse camera motion and pose annotations, we construct VG-Pairs with a synthetic paired-supervision data generation pipeline. Each video clip is treated as a geometry-consistent target and paired with a distorted counterpart synthesized by a pretrained image-to-video inpainting model using trajectory-adaptive keyframes. VGC learns to map the distorted input to the geometry-consistent target without modifying the source generator, and uses optical flow as an auxiliary cue to help disentangle camera motion from object motion. Experiments on pose-conditioned and prompt-driven benchmarks show that VGC improves geometric consistency across multiple source generators, and scaling to a larger backbone further extends the framework toward more general video refinement.

0.56

MultiAgent-OPD: End-to-End Training of Hierarchical Multi-Agent Systems with On-Policy Distillation

cs.CL ·2026-10-16 ·Yuhan Sun, Jiaxin Zhao, Yongdong Luo, Yibo Hu, Meiguang Jin +4 more
Abstract

In a hierarchical multi-agent system, a Leader assigns tasks and subagents carry them out. Jointly training the Leader and the subagents is difficult because a final answer gives only indirect feedback to the subagents. An LLM judge can score the performance of the subagents, but these scores depend on the judge's own assessment of the quality of subagents' responses. We present MultiAgent-OPD, which combines on-policy distillation with reinforcement learning to jointly train a shared-parameter Leader-Searcher-Answerer team end to end. We first train a Teacher in a Searcher-Answerer workflow using answer correctness. During team training, the frozen Teacher scores student-generated search and summary tokens to provide local process supervision for the Searchers, while the Leader learns from team success and the Answerer from candidate correctness. Across seven datasets, the trained system reaches 38.68% macro avg@4, exceeding Prompted Critic by 4.06 percentage points and C-MAS + RL, the trained cascaded multiagent system, by 2.61 percentage points. The team also solves more questions within four attempts.

0.54

DLite: History-free Speculative Decoding for Agentic AI

cs.CL ·2026-10-16 ·Jiaming Xu, Hanzhen Wang, Tianlang Zhao, Jiewen Xiao, Chengze Yuan +5 more
community pick
Abstract · 14 pages, 5 figures

Speculative decoding has become a mainstream technique for accelerating large language model (LLM) inference by using a lightweight draft model to generate candidate tokens that are subsequently verified by the target model. However, agentic workloads introduce long and continuously evolving contexts and frequent multi-turn interactions. Under such workloads, we observe that existing draft models suffer from up to an 80% degradation in acceptance length, while maintaining their historical KV cache introduces substantial system overhead. In this paper, we identify the growing historical context as a common source of both problems and introduce DLite, a history-free speculative decoding framework for agentic AI. DLite directly reuses contextual hidden states from the target model and performs local speculative prediction without explicitly maintaining the historical context, transforming an increasingly complex modeling problem into local simple prediction problem. A lightweight sequential head further restores intra-block causal dependencies with low overhead. To effectively train this heterogeneous history-free architecture, we introduce a tailored two-stage training strategy. Experiments on general-purpose, long-context and agentic workloads show that DLite achieves 2-6 acceptance length and outperforms the state-of-the-art speculative decoding method, DSpark, up to 70%.

0.53

OR-License: Evaluating LLM Agents as Real-World Operations Research Practitioners

cs.CL cs.AI ·2026-10-16 ·Peisong Wang, Feixue Liu, Yize Meng, Bowen Liu, Ruoyu Chen +14 more
benchmarkcode/datacommunity pick
Abstract · Code and Data: https://github.com/NineAbyss/OR-License

Can large language model (LLM) agents deliver expert-level operations research (OR) solutions directly from original client materials, without prior expert reformulation? Existing OR benchmarks typically begin with self-contained descriptions or predefined optimization tasks, underrepresenting the translation of original business materials into professionally validated solutions. We introduce OR-License, an end-to-end benchmark of 31 instances across 10 industrial scenarios from paid OR engagements totaling USD 3.68 million. It preserves original requirements, input data, deliverable formats, and information boundaries after de-identification. To study information acquisition, we introduce ORSim, a controlled client simulator that discloses business facts without supplying OR expertise. Hidden evaluation checks feasibility and compares objective quality against original expert solutions accepted by clients before benchmark construction. Deployment-Adjusted Quality (DAQ) averages expert-relative quality across instances, assigning zero to infeasible deliveries. Across 11 frontier LLMs using coding-agent harnesses, several exceed the expert reference in conditional objective quality, but none matches its DAQ; the strongest primary Claude Code configuration achieves 88.39% strict feasibility and a DAQ of 0.87 against the expert reference of 1. Client information acquisition extends beyond clarification into solution development and validation, with some later stages showing higher confirmed issue-resolution proportions. Trajectory analysis identifies formulation and implementation failures, while task-equalized research-chain completion rates range from 24% to 45% across models on evaluable tasks. These findings motivate sustained information-gap resolution, faithful problem formulation, and evidence-linked self-improvement for reliable long-horizon agents.

0.53

MOSS-Image Technical Report

cs.CV ·2026-10-16 ·MOSI. AI
code/datahot directioncommunity pick
Abstract · 35 pages. GitHub: https://github.com/OpenMOSS/MOSS-Image

Recent closed-source image generation systems can produce diverse visual content from simple compositions to information-dense graphics, yet open-weight text-to-image models still struggle with dense text and intricate layouts. To narrow this gap, we introduce MOSS-Image, an open generation system that combines high-fidelity visual synthesis with intent-aware composition. First, we pretrain a 5B single-stream diffusion transformer (DiT) from scratch with a carefully designed recipe, enabling bilingual long-form text rendering and precise layout control. Second, we develop an intention-aware agentic prompt enhancement (PE) module that translates diverse user requests into detailed content and layout specifications. Together, they enable MOSS-Image to generate text-rich and layout-coherent images. We further introduce LongLongTextBench (LLTB) for evaluating long-form English and Chinese text rendering, and RealLayoutArena (RLA) for end-to-end layout-rich generation. MOSS-Image achieves state-of-the-art performance among open-weight models on LLTB, with an average text coverage of 96.3%. On RLA, MOSS-Image ranks second in the end-to-end system comparison, while its backbone ranks first among generators evaluated with shared enhanced prompts. We will release the model, training recipe, and evaluation framework to support further research and promote the text-to-image community.

0.53

Persistent Planar Memory for Video World Models

cs.CV ·2026-10-16 ·Yuze He, Bin Tan, Zelin Gao, Yujun Shen, Yong-Jin Liu +2 more
top venuecode/data
Abstract · 20 pages, 9 figures. Accepted by NeurIPS 2026

Video world models that generate and explore environments at scale need a persistent 3D memory of what they have produced. Plane primitives offer a natural candidate: a compact set of oriented planes captures the dominant structural elements (e.g., walls, floors, facades) of indoor and outdoor scenes, while requiring only 7 parameters per primitive. We present PPMem, a video world model that adopts plane primitives as its persistent 3D memory. PPMem augments a video diffusion transformer with a planar prediction head and a Depth Token Fusion (DTF) module: each autoregressive chunk jointly produces RGB video and planar geometry, which merges into a persistent plane memory that conditions subsequent chunks via rendered depth. The resulting memory is three orders of magnitude more compact than per-pixel point clouds (~100K primitives, 5MB for a full scene), geometrically explicit (metric depth, normals, exportable surfaces), cross-view consistent (jointly predicted by the shared backbone), and adaptive (refined by the merging procedure as generation proceeds). On DL3DV and RealEstate10K, PPMem outperforms prior 3D-grounded world models in video quality while jointly producing metric-scale depth, surface normals, and exportable planar geometry, sustaining coherence across hundreds of frames under large viewpoint changes. Project page: https://hyzcluster.github.io/ppmem.

0.52

LUMINR: Intensity-conditioned Meta-learned INRs for Accelerated Medical Image Registration

eess.IV cs.CV ·2026-10-16 ·Damian Rynczak, Marek Wodzinski
hot direction
Abstract · 13 pages, 7 figures, 3 tables

Implicit neural representations (INRs) have been applied to deformable image registration, where a lightweight network maps a spatial coordinate to a deformation vector. Because an INR encodes one specific signal in its parameters, a fresh network must be optimized for every image pair at test time, which is slow. Existing methods reduce this cost by meta-learning an initialization across a population of image pairs, and condition the weights of the INR globally, so the same conditioning applies at every coordinate while the deformation field it initializes varies across the domain. We propose LUMINR, a meta-learning framework that accelerates this per-pair fitting and in which the conditioning is local instead. LUMINR uses an INR that maps a coordinate, together with local intensity and gradient features, to a deformation vector, so the conditioning varies over the domain at the resolution of the field being modelled. Its objective is to find an initialization of the INR parameters from which the network can be fine-tuned to register any unseen pair in only a few optimization iterations. We evaluate LUMINR on 3D deformable registration of brain MRI from the OASIS dataset, where local conditioning and the meta-learned initialization prove complementary: early in fitting they reach a markedly higher accuracy than either does alone. Fine-tuned from that initialization, LUMINR matches the accuracy of the same conditioned INR fit from scratch in roughly a tenth of the gradient steps and a tenth of the wall-clock time, while the deformation fields stay smooth with negligible folding.

0.51

Learning to Discriminate Scene Structures Makes Self-Supervised Depth Learning Scalable

cs.CV ·2026-10-16 ·Mengtan Zhang, Yuchuan Cui, Yu Ma, Zizhan Guo, Wei Ye +1 more
top venuecode/data
Abstract · Accepted to NeurIPS 2026

Self-supervised monocular depth estimation is appealing for its potential to scale with unlabeled data, yet existing methods often degrade when trained on diverse scenes. This study finds that, under self-supervision, models tend to encode dataset-specific structural biases rather than transferable scene geometry, leading to cross-domain optimization conflicts. To address this issue, this study proposes the MaSS framework, aiming to make self-supervised depth learning scalable to diverse data. MaSS introduces a dedicated structural representation pathway explicitly decoupled from contextual features, which is guided by a novel scene-structure discrimination objective to learn generalized, layout-aware structural representations for depth decoding. Trained and evaluated on four diverse datasets, both individually and jointly, MaSS substantially outperforms prior self-supervised methods and is the first to consistently benefit from mixed-domain training. Zero-shot evaluation on eight additional datasets further demonstrates its remarkable generalizability, particularly on those entirely out-of-distribution scenes. The source code is publicly available at https://mias.group/MaSS.

0.51

Self-Evolving Coding Agents via Harness-Policy Optimization

cs.CL cs.AI ·2026-10-16 ·Zhepei Wei, Wenlin Yao, Baolin Peng, Danqing Wang, Qianhui Wu +3 more
community pick
Abstract

Large language model (LLM) agents have shown promising potential in coding tasks, and their performance is jointly determined by the policy model and the harness that mediates their interaction with the environment. Indeed, harness and policy are mutually reinforcing and evolving only one component is suboptimal---a harness tuned for one checkpoint may not target the failure modes of an updated policy, while policy learning remains limited to the trajectories elicited by a fixed harness. In light of this, we propose Self-HPO, a recursive self-improving framework for coding agents that alternates harness and policy optimization, using the evolving policy as its own harness proposer. Specifically, in each round, the current policy first analyzes its own failures and proposes new harness candidates (H-step). The policy is then updated through reinforcement learning (RL) under the selected harness (P-step), and hard tasks identified during training form the next harness search set, keeping harness optimization targeted to the policy's evolving failure modes. Across three coding benchmarks, Self-HPO consistently outperforms harness-only and policy-only baselines, improving Qwen3.5-35B-A3B's average resolve rate from 57.8% to 64.8%, with gains generalizing to out-of-distribution settings. Further analysis shows that both optimized components retain gains even when paired with their base counterparts, while their joint gain exceeds the sum of their individual gains. Self-HPO also improves reliability (Pass@1) and expands the capability frontier (Pass@k) over the base agent. Additionally, we show that with a stronger external model (e.g., Opus 4.8) as the harness proposer, the same framework yields a further performance boost.

0.50

PCDFusion: Proposal-Context-Detail Bayesian Rendering for Infrared-Visible Image Fusion

cs.CV ·2026-10-16 ·Rui Liu, Lin Gu, Hesong Li, Ying Fu
publishedcode/data
Abstract · Accepted by NIPS 2026

Infrared-visible image fusion (IVIF) aims to combine visible and thermal infrared observations for human perception and downstream vision tasks. Visible and thermal infrared observations are physically complementary but unreliable in different ways. Existing deep IVIF methods mainly improve feature interaction and reconstruction, but often lack an explicit local rule for deciding when visible appearance should dominate and when thermal responses should contribute. To address this issue, we reformulate IVIF as local Bayesian posterior rendering, where fusion is controlled by posterior action weights rather than direct modality mixing. In this paper, we propose Proposal-Context-Detail Fusion (PCDFusion), a visible-anchored framework that implements this formulation through three rendering stages. PCDFusion preserves potentially useful thermal responses, updates exposure-degraded visible regions, and renders reliable band-limited infrared structures into the final luminance. Extensive experiments on public IVIF benchmarks show that PCDFusion improves fusion quality, target recovery, and downstream detection and segmentation performance under challenging illumination. The code is available at https://github.com/BIT-RuiLiu/PCDFusion.git.

0.49

Execution Before Memory: Diagnosing Memory Failures in Video World Models

cs.CV cs.CL ·2026-10-16 ·Weiyang Jin, Jiwen Yu, Sihan Xu, Yuwei Niu, Xihui Liu
code/datacommunity pick
Abstract · 27 pages, 7 figures. Project page: https://hku-mmlab.github.io/MIST/ Code: https://github.com/HKU-MMLab/MIST

A memory test for a video world model should say whether the scene was set, whether the asked change was carried out, and whether the last frame shows the result. Current tests fold these into one number, so a low score is easy to read as forgetting. When the change is not carried out, the hidden state transition cannot appear, so the test is not really a test of memory. We introduce MIST, an evaluation that separates the change from the memory it should leave behind. The model first sets the scene while the change stays hidden, and only later is asked to show it, with leakage before that point kept under control. This separation locates the bottleneck: the model sets the scene but does not carry out the instructed change. The same gap appears in general memory evaluations. On that basis we propose M3-Agent (Make the Missing Move), which recursively revises the details of the input instruction so that the model successfully carries out the requested instruction. Relative to the baseline, it more than triples the MIST reveal score. On the benchmarks, WRBench's score T rises by 26.1% and MBench-T's score U by 52.7%. The method meets the problem from the execution side: it shows what the frozen model can do once the move is stated, and it sets a stricter baseline.

0.48

Prefix-Tuning for Arbitrary Output Sequences on Pretrained Transformers

cs.CL cs.AIcs.LG ·2026-10-16 ·Peter Cho-Ho Lam, Ziyi Wang, Zirui Zhou
top venuetheorycode/dataunusual
Abstract · Accepted at NeurIPS 2026. Code: https://github.com/wzii/prefix_tuning. Online demo: https://wzii.github.io/prefix_tuning/

Prefix-tuning is a parameter-efficient fine-tuning method for large language models (LLMs) that steers model outputs by prepending task-specific prefix representations to the input sequence. Despite its practical success, the theoretical understanding of the expressiveness of prefixes remains incomplete. In particular, the following question remains open: Given a pretrained autoregressive Transformer, for any input-output sequence pair $(\mathtt{x},\mathtt{y})$, does there always exist a prefix $\mathcal{S}$ such that prepending $\mathcal{S}$ to $\mathtt{x}$ induces the model to generate the target output $\mathtt{y}$? In this work, we prove that under the assumption that $\mathtt{y}$ is in the range of the pretrained Transformer, such a prefix $\mathcal{S}$ exists and can be explicitly computed by solving multiple linear systems. We further show that this assumption is necessary, in the sense that such a prefix may not exist when the assumption fails, thereby providing a complete answer to the above question. Our analysis is constructive and the obtained theoretical results hold under fairly general assumptions on model architecture. Extensive experiments on GPT-2 and Qwen3-4B demonstrate that our algorithm produces prefixes with strong control over model outputs.

0.48

Think More or Read More? A Token-Budget Theory of Reasoning with Shared Context

cs.CL cs.LG ·2026-10-16 ·Manh Nguyen, Sunil Gupta, Hung Le
Abstract

Reasoning systems increasingly improve answers by incorporating additional context into the model's input, from multi-agent debate, where agents read their peers' responses, to self-refinement, where a model revisits its own drafts. In these systems, communication consumes context as one agent's output becomes another's input. We study the regime where token reading and generation draw on one fixed token window, so every token spent reading is a token that cannot be spent reasoning. We develop a \textbf{token-budget theory} that formalizes this trade-off, modeling accuracy as a function of the reasoning budget remaining after reading and aggregating votes with a Condorcet jury model that permits correlated errors. For multi-agent debate, the theory gives a closed-form prediction for the optimal number of agents. This quantity is estimated from single-agent measurements, allowing the optimal group size to be determined without running debate, and the same accounting extends to a heterogeneous set of different models. We empirically verify our theory and show that, at the predicted group size, our method outperforms strong baselines under the tightest calibration spend, where debate-calibrated rules can afford only one or two questions. Self-refinement obeys the same accounting, with its own draft in place of a peer message. The theory further predicts that communication pays only when the gain in the accuracy ceiling exceeds the reasoning room spent on reading, with this condition tightening as the window shrinks. Across five open models and four reasoning tasks, self-consistency matches or beats deployed debate at $93$\% of $980$ matched window-and-group-size settings, with the gap being largest under tight windows and nearly disappearing at a loose one, consistent with the theory. Under a shared window, how much room a model has left to \textit{think matters more than what it reads}.

0.48

NLData2Opt: Benchmarking Parameter Derivation and Prediction for Modeling Optimization Problems from Natural Language and Data

cs.CL ·2026-10-16 ·Lijia Liu, Kyohei Atarashi, Jiyi Li, Koh Takeuchi, Shunji Umetani +1 more
top venuebenchmarkcode/data
Abstract

Formulating optimization problems from raw inputs of problem descriptions and data is time-consuming and costly. While Large Language Models (LLMs) offer a promising way to automate this process, existing benchmarks overestimate their capabilities by relying on idealized inputs where all data is relevant and complete. This diverges from reality, where data-centric reasoning is required to extract, compute, and predict parameters from noisy sources. To address this, we introduce NLData2Opt, a benchmark to evaluate LLMs on the multi-stage reasoning inherent in practical optimization workflows. NLData2Opt decomposes problems into problem descriptions and data, and requires models to (1) extract relevant parameters from data containing misleading attributes, (2) compute missing parameters derived by transforming existing data attributes, and (3) predict unavailable future values from historical observations. Evaluation results show substantial performance decline relative to existing benchmarks, indicating that benchmark progress does not directly translate to formulation accuracy in practical optimization workflows. NLData2Opt is available at https://github.com/sumarivyn/NLData2Opt.

0.48

Kingfisher: Fast Personalized Visual Autoregressive Generation via Scale-Aware Prompt-Conditioned Scale Decomposition

cs.CV ·2026-10-16 ·Dinh-Khoi Vo, Akihiro Sugimoto, Minh-Triet Tran, Trung-Nghia Le
Abstract

Personalized image generation has been extensively studied in diffusion models, often relying on subject-specific optimization of model weights, textual embeddings, LoRA modules, or adapters. In this work, we explore a different direction for fast personalization in visual autoregressive models. Our key observation is that a short prefix of early autoregressive scale codes is sufficient to strongly determine both object structure and global appearance, while later scales primarily refine the established generation trajectory. Based on this insight, we introduce Kingfisher, a model-tuning-free framework that decomposes early residual scale codes into content- and style-related components through prompt-conditioned scale-code subspaces. Given two real images, Kingfisher first decomposes their early scale codes, then recomposes the content components of one image with the style components of the other into a compact semantic prefix. The frozen generator subsequently predicts the remaining scales through bitwise self-correction for coherent refinement. Experiments on CSD-100 show that Kingfisher achieves strong content preservation and the highest style and text alignment among the evaluated methods, while reducing pair-specific preparation time by approximately 54x compared with diffusion-based methods and 7.5x compared with VAR-based alternatives. These results demonstrate that effective personalization can be achieved by directly manipulating the native scale-space representation of a frozen visual autoregressive generator.

0.47

Jarvis: Coordination Deserves a Brain of Its Own in Agentic Interactive Models

cs.SD ·2026-10-16 ·Yuxiang Wang, Shengbo Cai, Zhixian Zhao, Yingda Shen, Yuancheng Wang +5 more
benchmark
Abstract

Interactive models increasingly hand complex reasoning and tool use to a slow brain in the background, leaving a realtime front model free to sustain full-duplex dialogue. Yet the user may refine, cancel or replace a request while its task runs, so a plan or result that made sense a moment ago goes stale. Each new utterance must therefore be read against the work in flight, but the front model follows only the conversation and the back end sees only its own task. Coordination deserves a brain of its own, one that sees both. We introduce Jarvis, a coordination harness where models judge and code enforces. Its typed decision model, JEV, reads each turn against the live tasks and picks among operations built from a versioned task record. Code checks every choice against the current state before applying it, so no decision can act on a task version the user has replaced. Jarvis also rechecks plans before they execute and results before they are spoken, and counts an answer as delivered only once the user has heard it in full. We build two instances of this design. Jarvis-Audio runs unchanged on three full-duplex speech models. Jarvis-Omni adds vision and rechecks the evidence behind an answer before and during playback, since a scene can change while nobody speaks. With the front model and back end held fixed, replacing Qwen Audio Agent with Jarvis raises tau-Voice task success more than sixfold, from 29 to 192 of 278 tasks, and the gains hold on every benchmark and front model. Against published systems, Jarvis-Audio reaches 69% Pass1 on Full-Duplex-Bench v3 with an open 9B front model, versus 60% for GPT-Realtime. These results suggest that how the two brains are coordinated matters as much as how strong they are.

0.46

The Window Is in the Reasoning: When LLM Agents Decide to Follow Prompt Injections

cs.CL cs.AIcs.CR ·2026-10-16 ·Haoxuan Jia, Yang Liu, Bin Chong, Yingguang Yang, Yancheng Chen +11 more
Abstract

LLM agents that read untrusted content can be hijacked by prompt injections. Runtime monitors that try to catch a hijack before the agent acts, and benchmarks that score each task with a single run, share an assumption: whether an agent follows an injection is settled when the injection is read. We test this assumption with branching rollouts that replay an episode exactly up to the step at which the agent reads an injection and then resample the rest of the episode many times, splitting the outcome variance into a part fixed at read time and a part decided afterwards. Across four AgentDojo suites, a Qwen3 agent without thinking is largely decided at read time (0.91 of the outcome variance lies between states); with thinking on, this share drops to 0.62 (95% CI of the drop [-0.38, -0.22]), in the same direction for three attack templates and two success labels. How much is fixed at read time is a property of the model that the same measurement reveals, and thinking lowers the attack success rate of Qwen3 but raises it for two other models. Resampling from reasoning prefixes shows that the decision forms gradually as the agent reasons. Zero-shot chain-of-thought monitors built on cheaper models miss much of what is already decided, whereas stronger monitors, and a lookahead monitor that judges simulated next actions, come close to what the prefix determines. Finally, single-run labels of thinking agents disagree with a rerun up to 35% of the time, so benchmarks need repeated runs per task.

0.46

Many Biases, One Circuit: Post-Training Tunes a Pretrained Cue-Routing Mechanism

cs.CL cs.AIcs.LG ·2026-10-16 ·Prakhar Gupta, Terry Jingchen Zhang, Zhijing Jin
Abstract

Language models often abandon a correct answer when the prompt contains a cue, a hint toward a wrong option such as "I think the answer is B." Sycophancy and related biases show this failure. They are often studied separately and blamed on alignment. We test this view causally, tracing the attention heads behind these biases through the stages of two post-training pipelines. In multiple-choice tests on most models, four cue types rely largely on the same few heads, a shared circuit already present in pretrained models. Most base models follow cues in plain text. Fine-tuning a base model on about 200 generic assistant examples without cues activates these heads in chat format, so cue-following can be a side effect of assistant tuning. The final reinforcement learning (RL) stage strengthens the effect of cues in almost every comparison. In Tulu-3-8B and OLMo-2-7B, we swap the attention patterns of six middle layers between the RL checkpoint and its predecessor. For suggested-answer cues, this shifts each checkpoint's cue effect most of the way to the other's level. Post-training thus seems to change when and how strongly models use this circuit. Mitigation for these multiple-choice biases could target this circuit.

0.46

CoDA: A Human-LM Collaborative Codebook Development Assistant for Domain-Specific Text Annotation

cs.CL cs.AI ·2026-10-16 ·Jaspreet Ranjit, Swaminathan Chellappa, Augusto Rivas Costante, Chandan Manjunath, Laura Onasch-Vera +3 more
top venuecode/data
Abstract · In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: System Demonstrations

We present CoDA, an open-source human-LM collaborative codebook development assistant that helps domain experts develop annotation guidelines for annotating new variables in unstructured text with language models. Existing language model annotation systems primarily focus on downstream classification performance, while assuming the availability of high-quality annotation guidelines, or codebooks. In practice, developing these codebooks is time-consuming and challenging, particularly for concepts that require expert knowledge. CoDA addresses this gap by collaborating with domain experts to iteratively develop codebooks through targeted review of incorrect language model predictions and incorporating expert feedback into the evolving codebook. We demonstrate CoDA on two real-world case studies: extracting health indicators from homelessness case management notes and classifying sentiment of wildlife trafficking posts on social media. Across both domains, CoDA produces codebooks that achieve downstream classification performance comparable to expert codebooks while reducing codebook development time from several weeks to two hours and requiring 5X fewer manually annotated examples. Applying CoDA to generate annotations at scale, we identified 2,410 case notes that contained health discussions or referrals and 10,196 social media posts expressing negative sentiment that may warrant investigation for potential wildlife trafficking. By accelerating codebook development, CoDA helps practitioners use language models to extract domain-specific insights from unstructured text at scale.

0.45

DiscoveryRSI: Recursively Learning to Discover Valuable Hypotheses

cs.CL ·2026-10-16 ·Jianyuan Zhong, Chenwei Wu, Zhaochen Yu, Xixian Jiang, Xuewei Li +18 more
code/data
Abstract · Code: https://github.com/Gen-Verse/DiscoveryRSI

Scientific discovery requires developing worthwhile hypotheses into practical investigations, yet learning from this evolving research process remains challenging. We introduce an operational definition of valuable and testable research hypotheses: evidence-grounded claims about consequential uncertainties, coupled with feasible investigations, explicit decision criteria, and alternative explanations. Our DiscoveryRSI framework develops hypothesis trees through multi-agent evolutionary search, compiles selected investigations into executable Harbor tasks, and recursively learns through three mechanisms: hypothesis credit optimization attributes scientific assessments to documented research contributions; harness replay identifies productive strategies and recurring failures to refine research workflows; and rediscovery applies accepted improvements to new investigations, generating further experience for subsequent learning. We also introduce DFM-Bench, which provides comprehensive assessment of hypothesis quality and cumulative research progress at the levels of hypotheses, revisions, development paths and research runs. Across four life-science domains, automated evaluation shows that hypothesis quality rises from 4.86 for the base model to 5.39 after two rounds of self-improvement, with the large gains in falsifiability and data resolvability. We further validate hypothesis quality through expert review, comparing our DiscoveryRSI-27B with other research agents. Preliminary human assessments show that DiscoveryRSI-27B scores higher than GPT-6 Luna on hypothesis importance and whether claims are plausible and evidence-bounded. DiscoveryRSI-27B achieved a 6.3% rate of hypothesis correspondence with research published after the model cutoff, largely outperform GPT-6 with 2.5% and Claude Opus 5 with 2.3%, while delivering greater human-rated improvement along research paths.

0.45

Ghost Attack: Using 19th-Century Illusions to Fool 21st-Century Biometrics

cs.CV ·2026-10-16 ·Tyler Fitzsimmons, Adam Czajka
publishedbenchmarkhot direction
Abstract · Accepted to WACV (Winter Conference on Applications for Computer Vision) 2027

While the Pepper's Ghost illusion is a 19th-century optical technique, we found that this forgotten gimmick exposes vulnerabilities in modern face recognition systems. This paper introduces the 'Ghost Attack': a physical attack that repurposes this classic illusion using low-cost, free-space hardware and a transparent medium to project high-resolution facial data into physical 3D space. We show in the paper that despite relying on centuries-old optical principles, this new attack type simultaneously breaks three components of modern facial recognition systems (FRS): (a) the face matcher, (b) the presentation attack detection module, and (c) the morphing attack detection component, resulting in high impersonation success rates against state-of-the-art algorithms without triggering presentation or morphing defenses. These findings reveal a significant blind spot in current biometric systems, which fail to account for transparent projections and evolving physical attack vectors. To support further research, we are offering the dataset of face images of 100 identities, re-captured with and without 'ghost' projections as a new benchmark representing a novel attack for face recognition.

0.44

TriINR: Unsupervised Atmospheric Turbulence Mitigation with Triple Implicit Neural Representations

cs.CV ·2026-10-16 ·Yuka Ogino, Rikuto Otsuka, Yuho Shoji, Takuya Ogawa, Takahiro Toizumi +1 more
published
Abstract · Accepted to WACV2027

Atmospheric turbulence degrades image quality in long-range imaging. The degradation is generally modeled with two spatio-temporally varying components, tilt and blur. Recent unsupervised turbulence mitigation methods have demonstrated the effectiveness of representing a sharp image as an implicit neural representation (INR). However, their tilt and blur representations remain frame-dependent, and the number of optimization parameters grows with the number of input frames. To address this limitation, we propose TriINR, an unsupervised turbulence mitigation method that recovers a sharp image from multiple degraded frames. TriINR represents the sharp image, tilt, and blur as three implicit neural representations. By modeling tilt and blur as spatio-temporal INRs, TriINR achieves frame-count-independent degradation representation and memory-efficient optimization. We further incorporate physical and statistical properties of atmospheric turbulence as constraints on the INRs to improve restoration accuracy. Experiments on synthesized turbulence data with three degradation strengths show that TriINR restores images more robustly across varying turbulence strengths than prior TM methods and achieves the highest accuracy among the evaluated unsupervised methods. Our method also demonstrates robustness on real long-range scenes containing additional non-ideal motion.

0.44

A Language-Informed Loss for Visual Generation

cs.CV cs.LG ·2026-10-16 ·Suchir Agarwal, Keshigeyan Chandrasegaran, Ziyu Chen, Kyle Sargent, Juan Carlos Niebles +2 more
code/data
Abstract · 29 pages; Project website: https://suchiragarwal.com/lilie

End-to-end visual generation with pixel-space loss has recently achieved competitive performance without the need for learned image tokenizers. In this work, we study whether language-informed supervision can provide a richer training signal than pixel loss for visual generation. We introduce LILIE (pronounced "lily"), a Language-Informed and Lossless Image Encoder. LILIE is an invertible image transform trained such that a vision-language model can recover captions from noised features. The encoder uses a normalizing flow transformer trained jointly with a vision-language model via next token prediction. We then use the encoder to extract a simple language-informed loss for text-to-image generation. We train JiT models with LILIE loss, reducing gFD by 14 points (66.84 $\to$ 53.01) against a pixel-only baseline on the GPIC dataset. Our code, encoders, and generative models are available at https://suchiragarwal.com/lilie.

0.43

CRAFT: Corrective and Robust Multi-Agent Framework for Text-to-Parametric CAD

cs.CV cs.AIcs.GR ·2026-10-16 ·Mohammed Musthafa Rafi, Anushrut Jignasu, Mahdi Saraeian, Chinmay Hegde, Aditya Balu +1 more
publishedcode/data
Abstract · Accepted at the 2026 IEEE International Conference on LLM-Aided Design (ICLAD). 11 pages, 5 figures, 10 tables. Code: https://github.com/ide…

Text-to-CAD generation has the potential to make mechanical CAD more accessible, but existing approaches face a trade-off between data dependence and output fidelity. Supervised methods rely on large paired text-model datasets, while zero-shot prompting strategies are often brittle and tend to lose the parametric structure needed for downstream editing and customization. We present CRAFT, a corrective and robust multi-agent framework that produces parametric OpenSCAD programs from natural language without task-specific training. At its core is a JSON intermediate representation that carries symbolic parametric expressions throughout the generation process, so the resulting models remain editable and retain their parametric structure. CRAFT combines semantic understanding, parametric planning, compilation, rendering, self-correction, and component-level verification within a layered recovery process, using multi-view visual feedback to detect and repair geometric errors. Across NopSCADlib, ABC, and Slice-100K, CRAFT is competitive with direct zero-shot baselines in perceptual and geometric quality while exposing substantially more editable parametric structure.

0.43

Bridging the Language Gap in Scholarly Data II: Enhancing Gender Inference for Chinese Names

cs.CL cs.CYcs.LGcs.SI+1 ·2026-10-16 ·Lisette Espín-Noboa, Mingrong She, Ana Maria Jaramillo, Liuhuaying Yang
code/data
Abstract · 36 pages (13 main; 23 SI), 6 figures in main (under review)

Perceived gender inference from names is important for inclusive scientometrics research, yet existing methods perform poorly on non-Western scholars. We address this gap for Chinese names, which appear in two forms, Hanzi and Pinyin, its Romanized form (e.g., Zhang Wei, Li Na). In international bibliographic records, names are often available only in Pinyin. However, a single pinyin name can correspond to multiple distinct hanzi characters with different gender associations, so gender inference from Pinyin alone is intrinsically ambiguous. To address this, we first develop two hanzi models, a regression model that produces a continuous gender score from 0 (male) to 1 (female), and a classification model that outputs one of three labels (\em{male}, \em{neutral}, \em{female}). We train them on a sample of hanzi names from Chinese-language physics publications, rated by 11 native Chinese speakers. On this sample, the models achieve an RMSE of $0.09$ and an R$^2$ of $0.85$ (regression), and macro-AUC of $0.97$ and a Cohen's $κ$ of $0.81$ (classification), demonstrating high performance that is robust to class imbalance. For pinyin names, we propose a pipeline that aggregates hanzi predictions via transliteration and frequency weighting, together with an ambiguity score that flags names whose gender cannot be resolved from Pinyin alone. Benchmarked against three off-the-shelf tools (Namsor, GPT, and Claude), our models achieve the highest agreement with human annotations for both gender scores and gender labels, in both scripts. Applied to an independent sample of Chinese-named authors in APS journals, our models assign a gender to every name, whereas existing Pinyin-based methods leave between 20\% and 100\% of names unclassified. These findings demonstrate the importance of hanzi-level information for reliable gender inference of Chinese names. Code and data are publicly available.

0.42

Direct Conditional Parameterization for N-Dimensional Splatting

cs.CV ·2026-10-16 ·Zhongpai Gao, Benjamin Planche, Van Nguyen Nguyen, Meng Zheng, Anwesa Choudhuri +3 more
top venuecode/data
Abstract · Accepted by NeurIPS 2026

N-dimensional splatting extends 3D Gaussian Splatting with view direction and time to model view-dependent and dynamic effects. Each primitive is lifted into a joint distribution over position and conditioning variables, then conditionally sliced at each rendering query. N-DGS and UBS share this step across Gaussian and Beta kernels: recovering the conditional mean, opacity, and covariance requires per-primitive matrix inversions and regression products in every forward pass. We introduce direct conditional parameterization, replacing this covariance-derived readout with explicit per-effect parameters: a Cholesky precision factor for opacity and a factorized query-to-position regression operator with learnable per-dimension coupling. Unrestricted conditional coordinates recover a valid joint covariance; our factorization adds spatial-scale and coupling priors. Applied to both kernels, the parameterization yields dGS (Gaussian) and dBS (Beta). Across five static and dynamic benchmarks, they match or exceed their covariance-derived counterparts on quality, with gains up to +1.26 dB in the main MCMC evaluation. Production slicing is 6.9-7.7 times faster than on-the-fly covariance-derived conditioning; complete-render profiling gives 1.25-1.91 times speedups on four trained scenes. Direct conditioning provides a structured, kernel-agnostic alternative for N-dimensional splatting. Project page: https://gaozhongpai.github.io/dgs/.

0.42

Chinese Decision: A Family of Language Models for Single-Pass Probabilistic Decision-Making

cs.CL ·2026-10-16 ·Qiang Yan
code/datahot direction
Abstract · 47 pages, 15 figures, 25 tables. Technical report v6. New since the previous arXiv version: unified 0.8B/2B/4B models with a typed span-extr…

Many applications need either a probability distribution over a few decisions or a set of typed spans from a text, and both are usually obtained by generating text. We present Chinese Decision, a family of Chinese models that answers both with a single forward pass and no generated token: candidate-letter logits give probabilities for Boolean, categorical, and ordinal decisions, and a small span head reading an echoed copy of the input gives typed character spans. Both readouts share one backbone and one merged LoRA, so a unified model is one weight file. We release decision-only and unified models at 0.8B, 2B, and 4B (Qwen3.5), trained for one epoch on 13.05 million decisions and, for the unified models, on CJ-IE, a 1.2-million-row Chinese extraction mixture. On the full 307,900-decision CJ-Bench, the unified 4B model reaches 86.86% accuracy, above every reported baseline and the decision-only 4B model (85.97%), and extracts spans on CJ-IE with micro F1 of 68.9. Retention has a price: against a 0.8B model continued on CJ-IE alone, which reaches 73.9 F1 but forgets CJ-Bench (71.02%), keeping decision ability costs 5.0 to 10.7 F1 points, shrinking with size. Gains of the unified models over decision-only ones are small, and all comparisons rest on single training runs with paired clustered bootstrap intervals.

0.41

Causal Latent Workspaces for Looped Transformers

cs.CL ·2026-10-16 ·Ilia Zaznov
code/data
Abstract · 20 pages, 5 figures. Code and reproducibility materials: https://github.com/izaznov/causal-latent-workspace-transformer

Reusing a Transformer block across depth increases computation without replicating its parameters, but successive passes have no dedicated state through which positions can exchange compact summaries of their causal prefix. We introduce the causal latent workspace Transformer: an anchored, four-round shared-depth architecture in which each position writes gated vectors into prefix-specific latent slots and retrieves a content-addressed summary before the next shared block application. This path complements causal self-attention while the original input embedding remains accessible at every round. Across three byte-level corpora, three seeds, and two matched parameter budgets, the proposed model achieves the best mean held-out perplexity in five of six settings against five Transformer controls. At 101.71M parameters it leads on WikiText-103 raw, enwik8, and TinyStories, reaching perplexities 2.7820, 2.7843, and 1.6823; these improve upon the strongest non-workspace control on each corpus by 1.06%, 0.73%, and 1.01%. All nine seed-labelled comparisons at this budget favor the workspace. At 0.918M parameters it leads on WikiText-103 and TinyStories, whereas a conditioned loop is stronger on enwik8. Architectural ablations show that adding the workspace improves upon the anchored loop in all six settings and support refreshing its causal summaries at each round. Together, these results identify a causal, content-addressable prefix workspace as an effective structural augmentation of looped Transformers.

0.40

SCALE: Improving Guidance-Free Visual Generation via Self-Contrastive Alignment for Likelihood Estimation

cs.CV ·2026-10-16 ·Jiwan Hur, Dong-Jae Lee, Dongyeun Lee, Gyojin Han, Junmo Kim
top venuecode/data
Abstract · Accepted to NeurIPS 2026. Project page: https://jiwanhur.github.io/SCALE/

Classifier-free guidance (CFG) is essential for improving sample quality in visual generation, but it incurs substantial sampling overhead. Existing guidance-free approaches halve this cost by approximating the CFG with a single model inference, but their performance is bounded by the CFG and inherits its limitations. In this paper, we propose self-contrastive guidance (SCG), which contrasts the model's prediction on real and self-generated samples to obtain a direct, model-based signal that pushes generation toward the true data manifold. Integrating SCG into the standard likelihood objective via reparameterization yields Self-Contrastive Alignment for Likelihood Estimation (SCALE), a guidance-free training framework that produces SCG-enhanced predictions in a single forward pass. Across class-conditional and text-to-image generation on diffusion and autoregressive models, SCALE consistently outperforms CFG-mimicking guidance-free baselines and improves further when combined with DPO, demonstrating that the self-contrastive signal yields gains complementary to both CFG and preference-based fine-tuning.

0.40

Self-Evolving Agents Can Repair Failures but Struggle to Generalize from Them

cs.CL ·2026-10-16 ·Weixiang Zhao, Yichen Zhang, Yanyan Zhao, Jinming Bai, Bin Liang +3 more
Abstract · 28 pages, 9 figures, 6 tables. Weixiang Zhao and Yichen Zhang contributed equally

Experience-driven self-evolution offers a promising path for large language model (LLM) agents to continually improve through interaction. Yet existing approaches often rely on successful trajectories or external evaluation signals, leaving unclear whether agents can evolve when such positive anchors are unavailable. We study this question through Cold-Start Failure-Only Evolution (CFOE), where agents start without accumulated experience and learn only from failed interactions without privileged guidance. Across three self-evolving paradigms, multiple LLM backbones, and four interactive environments, we uncover a consistent repair-generalization gap: failure-derived experience can help repair previously failed tasks, but rarely yields sustained improvements on unseen tasks. Further analysis shows that agents often misinterpret failures, while even verified-correct experience remains difficult to transfer. Relaxing CFOE reveals that verifier feedback and additional interaction provide limited gains, whereas successful trajectories substantially improve transfer. These findings suggest that current self-evolving agents are better at repairing observed failures than building generalizable capabilities from failure.

0.40

An Efficient Approach to Beetle Damage Detection of Banana Using Deep Learning

cs.CV cs.LG ·2026-10-16 ·Md. Mehedi Hasan Naeem, Ishtiyak Ahmmad Araf, Md. Arefin Haque Mahir, Tushar Kanti Saha
publishedhot direction
Abstract · 6 pages. Accepted at the 12th IEEE International Conference on Sustainable Technologies and Engineering (IEEE iCOSTE 2026), Cox's Bazar, Ban…

Visible surface damage of banana fruits by Basilepta subcostata (the banana fruit scarring beetle) can reduce their market value. Here, manual inspection of Banana fruits is subjective and difficult to scale. This study presents a lightweight framework based on YOLOv8n deep learning model for locating individual bananas and classifying their visible surface condition as healthy or unhealthy. The term beetle damage in the paper title is used operationally for the locally observed scar-like surface condition; the model does not identify a specific beetle species or biological cause. A custom dataset of 2,473 images containing 5,827 annotated banana instances was prepared from local-market imagery in Bangladesh. The reported experiment used an 80/20 protocol with 1,982 training images and 491 test images. A pretrained YOLOv8n model was trained at 640x640 resolution for 50 epochs with a batch size of 4. On the test partition, it achieved 81.36% precision, 74.30% recall, 77.67% F1-score, 87.04% mAP@0.50 and 59.33% mAP@0.50:0.95. A YOLOv8s comparison gave slightly lower F1 and mAP while using a substantially larger checkpoint. The YOLOv8n checkpoint occupied 5.96 MB and required 6.54 ms/image for inference on an NVIDIA Tesla T4. The results support YOLOv8n as a compact baseline for instance-level banana surface inspection within the studied acquisition conditions.

0.40

HAPACT: A Benchmark For Human-Centric Physical Impact Localization in Movies

cs.CV cs.AI ·2026-10-16 ·Youngrae Kim, Yejin Jang, Eunho Kim, Youjin Sung, Kun-Woo Song +2 more
top venuebenchmark
Abstract · 29 pages, 16 figures. Accepted at NeurIPS 2026 (Evaluations & Datasets Track)

Physical reasoning about humans in videos has largely focused on everyday interactions, where contact, pressure, pose, and motion support embodied perception and contact-aware applications. Still, short and forceful properties like impact events remain relatively underexplored, despite their relevance to broader understanding of how abrupt physical events affect the human body. To fill this gap, we introduce human-centric impact localization, a new task that localizes target physical impacts temporally within a video and spatially on the human body. We further present HAPACT, a HumAn-centric imPACT dataset and benchmark built from movie clips, containing over 2.8K video-query pairs with more than 20K fine-grained human-annotated frames. We built HAPACT around viewer-grounded annotation, text-conditioned localization, and supervision at frame-level temporal and SMPL-X-based spatial granularity. To demonstrate the effectiveness of our benchmark, we conduct extensive evaluations of existing methods, employing video moment retrieval models for temporal localization and human-scene contact models for spatial localization. For the spatial task, we propose a text-conditioned baseline by reformulating localization as query-guided body-region selection. Benchmark results show that physical impact localization remains challenging for existing methods, motivating HAPACT as a dedicated benchmark to support physical impact reasoning in videos.

0.39

Scene-Designer: Learning Design Intelligence for Agentic 3D Scene Generation

cs.CV ·2026-10-16 ·Xuanyu Liu, Hongyan Feng, Miao Pan, Sunlai Chen, Lizhou Liang +3 more
Abstract

Generating complete, coherent, and usable 3D indoor scenes from natural language requires strong \textbf{design intelligence}: the ability to design high-quality scenes and critique their quality. However, existing approaches either rely on general-purpose models lacking domain-specific design experience or learn from static scenes. In this work, we introduce \textbf{Scene-Designer}, an agentic framework that learns design intelligence through complementary design and critique experiences. We construct two datasets: \textbf{Scene-OTA-35K}, which provides Observation--Thought--Action supervision for learning scene design and efficient error correction, and \textbf{Scene-OTC-15K}, which provides Observation--Thought--Critic supervision for learning scene quality critique across physical validity, functional correctness, and layout coherence. Based on these experiences, we propose \textbf{Scene-GRPO}, a Critic-guided GRPO framework that improves design decisions and reduces unnecessary trial-and-error corrections. Extensive experiments demonstrate that Scene-Designer achieves superior performance in physical validity, functional correctness, and layout coherence, while the learned Scene-Critic significantly outperforms general-purpose models in scene critique. Moreover, test-time Critic guidance further improves scene generation, showing that the Critic can provide useful guidance beyond training. Additional analysis suggests a trend of mutual enhancement between scene design and critique abilities.

0.39

RAP: A Dataset for Paper-to-Poster Generation via Declarative Markup from Human-Crafted Exemplars

cs.CV ·2026-10-16 ·Haonan Han, Yukang Lin, Minliang Lin, Haoxian Tan, Jun-Yan He +5 more
benchmarkdatasetcommunity pick
Abstract

Making a conference poster is visual coding in miniature: an entire paper must be compressed onto a single page, which exercises summarization, visual design, and code implementation at once, using nothing beyond the paper itself. Yet benchmarks for this task in the context of visual coding remain scarce: existing resources are small and annotated only with layout boxes, so evaluation must fall back on VLM-as-judge or human checklists rather than objective render-and-compare, and the generation methods they support design layouts from scratch, leaving human-crafted posters unusable as either a retrievable design library or as verified supervision. We introduce RAP (Reconstruct A Poster), a dataset and benchmark for paper-to-poster generation as a visual coding task. RAP contains 25K+ verified paper-poster pairs from various machine learning conferences held between 2022 and 2026, annotated with a three-level content taxonomy, plus a held-out benchmark of 150 test papers each paired with its human poster. A fully automated derendering pipeline turns each poster into styled HTML/CSS of typed, refillable slots, so every sample is a paper-poster-markup triplet verifiable by rendering against the source poster. The corpus serves both inference and training: PosterRAG, a training-free multi-agent pipeline, retrieves an exemplar and refills its slots without touching layout or styling, while an SFT+RL recipe trains a dedicated model from the same triplets. An online demo is available at https://reconstruct-a-poster.github.io/.

0.39

KiranaBench: Can a Language Model Run a Neighbourhood Shop?

cs.CL cs.MA ·2026-10-16 ·Bhuvan Arora, Ankita Karmakar, Chaitanya Jain, Dilesh Churiwal, Harshit Jain +8 more
code/data
Abstract · 27 pages, 16 figures, 7 tables. Code, data and traces: https://osf.io/s2pxg/

Large language model (LLM) agents are increasingly evaluated on long-horizon tasks, including autonomously running a business. Existing benchmarks for this setting are almost exclusively English-language, Western in retail context, and single-agent, with the agent observing the business directly. We study how these results transfer to a multi-agent Indic setting. We introduce KiranaBench, a simulated Indian general store in which four roles (manager, cashier, helper, and bookkeeper) are each played by a separate LLM agent. Interactions with customers and suppliers occur in English, Hindi, or Hinglish, and product demand is calibrated to India's national Household Consumption Expenditure Survey. Evaluating four open-weight LLMs, we find that all fall far short of a fixed-rule shopkeeper baseline: the best model earns only one-eighth of the baseline's profit, and two models finish below an unmanaged shop. Performance is also consistently lower for Hindi and Hinglish than for English supplier communication. KiranaBench offers evidence that current LLM agents are not yet reliable for autonomously operating small businesses in India, and provides a testbed for developing agents suited to settings that are long-horizon, collaborative, and multilingual.

0.39

VocalGrad: Evaluating Acoustic Perception in Audio Language Models

cs.SD cs.CL ·2026-10-16 ·Koki Ryu, Hitomi Yanaka
publishedbenchmarkcode/data
Abstract · NeurIPS 2026 (Evaluations & Datasets Track). Dataset page: https://huggingface.co/datasets/ynklab/vocalgrad

Recent large audio language models (LALMs) have extended their capabilities beyond lexical speech understanding to tasks requiring paralinguistic and non-linguistic audio understanding, such as emotion recognition. However, these models still struggle to reliably utilize acoustic information, particularly when it co-occurs with spoken linguistic content. Existing benchmarks do not clearly distinguish acoustic perception from higher-level reasoning, making it difficult to identify the source of these failures. To address this issue, we introduce VocalGrad, a benchmark designed to evaluate LALMs' ability to perceive low-level acoustic variations. VocalGrad formulates this problem as a simple binary classification of temporal change direction, enabling evaluation largely independent of higher-level reasoning or semantic understanding. Our benchmark results reveal a substantial gap in acoustic perception between human annotators and state-of-the-art LALMs: while human annotators achieve high accuracy, LALMs perform at near-chance levels. We further conduct linear probing to identify structural bottlenecks in these models. Our analysis reveals two distinct failure modes: for some categories, task-relevant information is preserved in internal representations but not utilized by the language model head; for others, the information becomes progressively less linearly decodable across model layers. Overall, our results reveal substantial limitations of LALMs in modeling acoustic variation under controlled settings and expose representation differences not observable from text outputs.

0.39

From GAN-Based Models to GPT-5: A Taxonomy-Guided Study of Human Performance in Detecting Deepfake Face Images

cs.CV cs.CRcs.HC ·2026-10-16 ·Zhiqing Guo, Lixin Jia, Luning Sun, Xiaowei Chen, Guofeng Yu +5 more
publishedhot direction
Abstract · Accepted by IEEE S&P 2027

Generative models for deepfakes have rapidly evolved from GAN-based architectures to diffusion models and multimodal large language models (MLLMs). However, empirical research on human performance in detecting deepfake face images has not kept pace. It remains unclear if (and how) detection performance varies across generative models and whether model-specific training improves detection more effectively than general-purpose training. In this paper, we propose the \emph{Deepfake Face Artifact Taxonomy} (DFAT), a cross-model taxonomy of human-recognizable visual artifacts in deepfake face images, based on which we build both general-purpose and model-specific training materials. Through an online experiment ($n = 524$), we examined human performance in detecting deepfake face images and evaluated training efficacy across five representative generative models: three GAN-based, one diffusion-based, and one MLLM (GPT-5). Our results show that participants' detection accuracy was strongly model-dependent and that images generated by more recent and advanced models were not necessarily harder to detect. After training, participants achieved significant performance gains across most conditions, and their improvement varied by model. Notably, model-specific training generally yielded greater numerical gains than general-purpose training. In a five-month follow-up test ($n = 342$), we observed no significant decline in detection accuracy and no significant increase in perceived stress. Overall, our findings provide actionable insights for designing effective, sustainable, and targeted training programs to counter the growing threat posed by deepfakes.

0.38

CLEAR: Cooperative Latent Flow for Visual Counterfactual Explanations with Accurate Region Preservation

cs.CV ·2026-10-16 ·Yan Zeng, Changlu Guo, Oskar Kristoffersen, Anders Nymark Christensen, Morten Rieger Hannemose +1 more
publishedcode/data
Abstract · Accepted to WACV 2027

Visual counterfactual explanations (VCEs) should alter a classifier's prediction while remaining realistic and preserving decision-irrelevant content. Image-space methods provide high-quality counterfactuals but are computationally costly, whereas latent-space methods are more efficient but often less faithful. We identify two sources responsible for this gap. First, latent generation recovers a natural sample from a perturbed representation, whereas classifier guidance pushes it toward a target class, so their update directions may oppose. Second, VAE reconstruction can alter regions that should remain unchanged. We introduce CLEAR, a latent Flow Matching framework for VCEs. CLEAR uses Flow Matching's explicit instantaneous velocity as a reference for the generative direction to decompose classifier guidance into cooperative and conflicting components. It preserves the cooperative component and adaptively attenuates only the conflicting one according to conflict strength. We further analyze conflict frequency and strength throughout generation to reveal how classifier guidance interacts with the generative process. CLEAR also dynamically identifies editable regions, constrains latent modifications, and directly preserves factual content outside them to reduce VAE-induced changes. Across five bidirectional tasks on two datasets, CLEAR retains latent-space efficiency while achieving counterfactual quality comparable to or better than strong image-space baselines, with a 2.5x speedup over the strongest baseline. Project page: https://yan-zen9.github.io/CLEAR/

0.38

Xiaomi-Lens-0 Technical Report

cs.CV ·2026-10-16 ·Pei Fu, Yike Liu, Yue Sun, Hanmo Wang, Xiuwen Xi +11 more
community pick
Abstract

Multimodal large language models (MLLMs) answer questions about whole images well, yet they still struggle with fine-grained region understanding. A key reason is that text-only supervision never specifies which regions support an answer. To close this gap, we present Xiaomi-Lens-0, a 4B MLLM built on Qwen3.5 that predicts masks from its own visual representations. Specifically, a lightweight decoder combines vision-encoder patch features with the language model's mask-token states, termed visual latents, so that mask supervision reaches the shared backbone without an additional segmentation encoder. Because each latent decodes to a mask, the model can interleave reasoning with region-specific latents and thus think with visual latents, while grounded reasoning traces supervise the intermediate regions referenced during chain-of-thought. As a result, across all eight RefCOCO, RefCOCO+, and RefCOCOg splits, Xiaomi-Lens-0 gains 5.6 points on average over Qwen3.5 in referring comprehension and achieves the highest segmentation cIoU among compared models. Moreover, it surpasses Qwen3.5 across all perception benchmarks. Homepage: https://seerray-lab.github.io/Xiaomi-Lens-0

0.37

Humanity's Last Choice: A Common Evaluation Standard for Generative Engine Optimization

cs.IR ·2026-10-16 ·Qianfeng Wen, Yifan Simon Liu, Difan Jiao, Junda Wu, Di Bai +3 more
Abstract · 19 pages, 2 figures. Project website: https://humanitylastchoice.org

Humanity's Last Choice describes the idea that the last choice people need to make is which AI system they entrust with the decisions that follow. These systems rely on external sources that content providers can edit through generative engine optimization (GEO) to increase source visibility and influence recommendations. Choosing among these systems therefore requires comparable evidence of how well they carry out user requests when exposed to GEO. Benchmarks have made progress in standardizing tasks, models, and metrics, but they often define their own spaces of GEO strategies and use different editing constraints. These differences can conflate gains from better optimization algorithms with those from greater freedom to edit sources, while making evaluation coverage difficult to compare across benchmarks. We therefore call for a common evaluation standard and propose a meta-taxonomy that describes where sources are edited, how they are edited, and what signals those edits target. Drawing on representative literature, we map existing strategies into a searchable space and outline a protocol for controlled comparisons with explicit coverage reporting. This provides a foundation for common leaderboards that can inform which systems people choose to rely on.

0.37

Detection of false positives by human reviewers depends on the face recognition algorithm and its security configuration

cs.CV cs.HC ·2026-10-16 ·Esha Edakoth, John J. Howard, Isabelle M. Shuggi, Yevgeniy B. Sirotin, Robert Wood
Abstract · 26 pages, 8 figures, 4 tables

Integration of artificial intelligence (AI) technology into identity workflows, such as automated face recognition (FR), has enhanced operational capabilities, with FR algorithms achieving super-human scale and accuracy. However, these gains can make human oversight of FR more difficult, raising questions about the effectiveness of these human-AI teams. We measured human reviewers' ability to detect false positive errors made by commercial FR algorithms across varying security configurations. Using a novel dataset, the Difficult Facial Image Comparison Test (DFICT), derived from false positives made by four comparable commercial face recognition algorithms, we measured the ability of 801human reviewers to detect algorithmic false positives across three security configurations, representing operationally relevant biometric thresholds. Our findings show that human reviewers can reduce algorithmic false positives, but performance declines at stricter security thresholds. Further, reviewer performance varied by algorithm indicating that some algorithms are more compatible with human oversight. We suggest that biometric systems should be designed to account for human reviewers' variable effectiveness, particularly in high-security applications. Our findings should be considered during algorithm selection and inform human oversight strategies to minimize AI-related risks in operational FR deployments.

0.37

Can the Parts Fool the Test? Counterfactual Pairing Cycles for Relational OOD

cs.CV cs.LG ·2026-10-16 ·Yibo Zhou, Yirong Xiang
publishedbenchmarkcode/data
Abstract · Accepted at Neurips 2026

Relational out-of-distribution errors are easy to describe and hard to evaluate cleanly. An image may be natural, a caption may be fluent, and all mentioned objects may be common, yet the caption may describe the wrong relation in the image. Many evaluations intended to test this behavior can be solved without checking the pairing at all, because negatives differ in caption style, template, length, object frequency, or image statistics. We propose counterfactual pairing cycles, a blocked evaluation primitive that reuses the same two images and the same two captions on both sides of the comparison, changing only which image is paired with which caption. This makes image-only, text-only, and additive component shortcuts cancel exactly in finite samples. The same contrast also gives a training loss for relational outlier exposure and a diagnostic for bilinear vision-language scores. Across synthetic controls, GQA, and COCO, cycles separate coarse image-caption matching from directional role-binding failures, while also identifying geometric size relations that remain strongly visible in global features. In a shortcut-contaminated COCO outlier-exposure stress test, a model selected by a near-perfect shortcut validation AUROC retains substantial relational signal but remains below cycle training on exact cycles; cycle training improves CycleAcc from 0.882 to 0.927. The main lesson is simple: to test whether a model understands a pairing, the benchmark should keep the parts fixed, change the pairing, and ask only about coupling. The official implementation is available online.

0.37

The Collection of a Real-World, Large-Scale Dataset for Automated Skill Evaluation in Central Line Dressing Change

cs.CV ·2026-10-16 ·Ummey Tanin, Shannan K. Hamlin, Hsin-Mei Chen, Peizhu, Qian
benchmark
Abstract · 9 pages

Central line-associated bloodstream infections (CLABSIs) remain a major source of preventable morbidity and mortality in healthcare settings. Proper execution of the central line dressing change (CLDC) procedure is essential for infection prevention, yet current approaches to competency assessment rely primarily on manual ob- servation by clinical educators, making evaluation time-intensive, difficult to scale, and limited in consistency and granularity. These challenges highlight the need for AI-assisted skill evaluation methods that can support more objective, reproducible, and scalable assessment of clinical performance. However, progress in this area has been constrained by the lack of broad-scale, clinically grounded datasets for developing and evaluating such methods. In this work, we present the collection process of a large-scale video dataset for CLDC skill evaluation, comprising 365 recordings of registered nurses collected during the 2026 annual competency exam at Houston Methodist Hospital. Each video is paired with subtask-level temporal annotations and clinical performance assessments spanning 17 subtasks of the procedure. We describe the data collection and curation pipeline, including pilot testing of record- ing configurations, annotation protocol design, and post-collection processing. We further establish the feasibility of video-based skill assessment through benchmark experiments with spatiotemporal deep learning models. This dataset provides a foundation for future research on automated clinical skill assessment, feedback generation, and AI-assisted training in nursing education.

0.37

Analyzing Occupational Gender Bias across the LLM Training Lifecycle

cs.CL ·2026-10-16 ·Manthan Gadhia, Alexander Hoyle, Karolina Stańczak
code/data
Abstract · Submitted to 7th Workshop on Gender Bias in Natural Language Processing @AACL-IJCNLP 2026

Post-training is widely assumed to reduce the social biases language models inherit from pretraining, yet direct evidence for this assumption remains mixed. In this paper, we examine how gender disparities evolve across the language modeling pipeline. In particular, we study occupational gender bias in OLMo 3, a fully open language model family, across its post-training stages by prompting each checkpoint to generate open-ended narratives about synthetic personas given their occupation and education. We consider two settings, a gender-given setting in which the profile states the persona's gender explicitly and a gender-assumed setting in which it must be inferred by the model. We find that when gender is stated, post-training increases semantic diversity for male-coded profiles, whereas for female-coded profiles the increase is smaller after SFT and not significant after DPO and RLVR. When gender is assumed by the model, post-training affects narratives about female and male personas similarly. The model's own gender assumptions, however, show a consistent female-preferential skew relative to United States labor-force demographics, except for heavily male-dominated occupations. Every successive post-training stage monotonically increases the odds of attributing a female label over the base model. A provenance analysis of the supervised fine-tuning corpus finds only a modest 7.7% female-token surplus, too small to explain the size of these downstream effects. The project code is available under: https://github.com/manthangadhia/llm-lifecycle-bias.

0.37

X-Splat: Unified Gaussian Splatting for Joint RGB and X-ray 3D Reconstruction

cs.CV cs.GR ·2026-10-16 ·Wangze Xu, Chu Zhou, Mae Yamaguchi, Linrui Dai, I-Chao Shen +2 more
publishedcode/data
Abstract · SIGGRAPH Asia 2026 Conference Papers. 19 pages (12 main paper + 7 supplementary), 13 figures. Project page: https://zezeaaa.github.io/X-Spla…

Jointly reconstructing RGB and X-ray provides new insights by connecting external appearance with internal structure, enabling a more complete understanding of real-world objects than single modality. Yet existing methods train separate models per modality and suffer from redundant geometry, alignment artifacts, and loose coupling that limits cross-modal insight discovery. We propose a unified representation in which all Gaussian primitives share a single set of geometric parameters, with only surface-proximal primitives additionally carrying appearance attributes. We perform cross-modal registration through clean multi-view stereo and X-ray geometric proxies instead of aligning two sets of optimized Gaussians, improving alignment stability. We then jointly optimize the unified representation via supervision from both modalities, complemented by a surface-depth consistency loss derived from X-ray geometry to guide RGB reconstruction. Experiments on the InsideOut dataset and our Flower & Insect dataset show that our framework achieves more robust cross-modal alignment, higher RGB rendering quality without compromising X-ray reconstruction, and a more compact representation than maintaining two independent geometries.

0.36

ADA 2025: A Multimodal Dataset for Water-Deficit Stress Detection in Sugar Beets

cs.CV ·2026-10-16 ·Piper Powell, Robert Franz, Amos Smith, Isaak Ihorst, Joachim Hertzberg +1 more
code/data
Abstract · 6 pages, 4 figures, 1 table. Accepted at the 2026 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor). Data…

Research and development into artificial intelligence (AI) and machine learning (ML) applications in crop management are hindered by a lack of suitable public datasets. Whereas applications in disease and pest detection benefit from a comparatively higher amount of available data, applications like water-deficit stress detection often have very few if any public datasets available. Moreover, most datasets that are available are single-modality and frequently RGB-only, leaving the early detection and enhanced performance offered by other methods, and particularly by multimodal approaches, out of reach. Datasets with accurate metadata, higher numbers of plants, and smaller intervals between data points are also scarce. To address this, we publicly release ADA 2025, a >500 GB comprehensive multimodal dataset of young sugar beet plants under healthy and water-deficit conditions combining thermal, hyperspectral, RGB, and depth imagery along with relevant environmental and metadata and both relative water content (RWC) and leaf thickness as ground truth sources. Exemplary results showcase the dataset's validity in terms of differences detectable between stressed and healthy plants.

0.36

Agentic Image Generation in the Wild

cs.CV ·2026-10-16 ·Yifan Li, Bowen Qin, Kun Zhou, Haochen Zhao, Zikang Liu +6 more
benchmarkcode/data
Abstract

Agentic image generation pairs an image generator with a search agent that retrieves references from the web, extending generation beyond the generator's training data. Real-world requests, however, often concern entities that appeared only days ago. We call this setting Image Generation in the Wild. In this setting, the web offers limited visual evidence, search results are crowded with easily confused entities, and the target is rarely shown on its own. To study this setting, we build an automated pipeline that synthesizes prompts from recent headlines and keeps only those whose real search results are highly confusing, yielding requests that are hard to solve yet natural to ask. With it, we construct Savanna, a benchmark of 500 prompts across 16 subcategories. On Savanna, even harness frameworks built on the frontier models retrieve a correct reference for only about half of the prompts. We then propose Locate-Corroborate-Isolate, a paradigm in which the agent first pins down the target's identity through text search, then verifies candidate images against multiple sources, and finally crops a clean reference for the generator. Trained on 15K trajectories that follow this paradigm, Ranger-8B achieves a search score of 67.2, outperforming harnesses equipped with frontier models. It also achieves the best generation quality with every generator we test, and surpasses prior agentic methods on other image generation benchmarks. We release the pipeline, benchmark, training data and model at https://github.com/RUCAIBox/Agentic-ImageGen-Wild.

0.36

ReLaDrift: Recurrent Language Generation via Drifting for Test-Time Scaling

cs.CL ·2026-10-16 ·Yuki Otsuka, Daisuke Oba, Naoaki Okazaki
code/data
Abstract

Drifting Models generate a sample with a single function evaluation, making them attractive for fast parallel language generation, but their one-step formulation does not support iterative refinement. We propose Recurrent Language Generation via Drifting (ReLaDrift), which uses a shared Drifting Model to generate an initial draft from noise and iteratively refine it in a continuous latent space. We train this recurrence by sampling the number of rounds and applying the drifting loss only at the final round, stopping gradients through preceding rounds so that only one round requires backpropagation. Unlike the one-step Drifting Model, whose accuracy does not improve when iterated, ReLaDrift enables a trade-off between test-time computation and generation quality, as diffusion and flow language models do. On Sudoku Hard and GSM8K, it achieves competitive or superior performance compared with these iterative-generation approaches across a range of inference budgets. Our code is available at https://github.com/Yuki-Ots/ReLaDrift

0.36

Who Pays for Efficient Inference? Confidence-Based Small-to-Large LLM Routing Across Tasks and Languages

cs.CL ·2026-10-16 ·Md Sabbir Hossen, Anichur Rahman, Pabon Shaha, Airin Afroj Aishi, Andrew H. Sung +1 more
theorycode/data
Abstract

Small-to-large language model routing can reduce inference cost by escalating difficult inputs while retaining easier ones on a smaller model. Its success, however, depends on identifying examples that benefit from escalation, and the reliability of confidence across multilingual tasks remains unclear. We present a systematic evaluation of confidence-based routing across three model families, four languages, and three task types, covering 72 model, dataset, language, and scale configurations and 42,288 predictions. We compare global confidence routing with random routing, equal-quota confidence routing, and an oracle upper bound. Confidence identifies beneficial escalations on the Belebele reading comprehension and SIB-200 topic classification datasets, with mean correction-detection AUCs of 0.612 and 0.705. At 50% escalation, it improves over random routing by 1.40--4.67 percentage points on Belebele and 2.54--5.63 points on SIB-200. In contrast, the mean AUC on the MMLU-ProX-Lite knowledge and reasoning dataset is only 0.510, and routing provides no consistent improvement over random selection because small models often remain highly confident while incorrect. Global confidence routing also creates language escalation gaps up to 30.33 percentage points, whereas equal-quota routing removes this disparity with little aggregate accuracy cost. These results show that confidence routing is task-dependent and can allocate computation unevenly across languages. Code is available at https://github.com/sabbir5622r/multilingual-llm-routing

0.36

Trajectory-Triggered Sparse Collaborative Decoding

cs.CL cs.LG ·2026-10-16 ·Amit Ranjan Trivedi, Sruthikesh Surineni, Kyong Ho Lee, Mohit Mittal, Steven Li +1 more
Abstract · 17 pages. Accepted at the EMNLP 2026 uncertainty workshop

Dense assistance methods such as speculative decoding repeatedly verify draft tokens with a target model, keeping target computation coupled to generation. We study sparse collaborative decoding, where a draft model generates most tokens and a target model is invoked only for short corrective segments. The key problem is deciding when to call the target. We show that local uncertainty is a weak trigger: entropy, negative log-likelihood, logit margin, top-$k$ mass, and velocity features give near-chance failure prediction, with AUC 0.50-0.53. In contrast, aggregate trajectory-shape features over the draft confidence path reach AUC 0.86 using a lightweight logistic predictor trained only from draft correctness labels. This induces a trigger-timing tradeoff: early calls have more leverage but weaker evidence, while later calls have stronger evidence but less ability to redirect the trajectory. Used online, learned sparse collaborative decoding improves over the best open-loop schedule by +5.3 to +20.0 accuracy points across five GSM8K draft-target pairs. On MATH-500, it gains 7.0-13.5 points when early prediction and target competence align, but loses when either is weak. On StrategyQA, it improves over open-loop by 5.0-8.3 points, showing transfer beyond arithmetics.

0.36

Colour-Field Based Particle Categorization for Residual Stress Detection and Reduction in Solid SPH Simulations

cs.GR cond-mat.mtrl-sciphysics.flu-dyn ·2026-10-16 ·Gizem Kayar
publishedunusual
Abstract

Residual stress remains in an object even in the absence of external forces or thermal pressure, which, in turn, may cause significant plastic deformations. In case the residual stress creates unwanted effects on the material and so is undesirable, an efficient solution is necessary to track and eliminate this stress. Smoothed Particle Hydrodynamics has been extensively used in solid mechanics simulations and the inherent colour-field generation approach is a promising tracker for the residual stress. In this paper, we propose a way to use the colour-field approach for eliminating the residual stress and prevent the undesirable premature failure of solid objects.

0.36

MAS-Bai: Linguistically Informed Adaptive Multi-Agent Translation for Bowen-Script Bai

cs.CL ·2026-10-16 ·Jingting Zheng, Hong Ren, Yuehan Cui, Shaolin Zhu, Deyi Xiong
top venue
Abstract · 14 pages, 3 figures, Findings of EMNLP 2026

Large Language Models (LLMs) remain unreliable for extremely low-resource languages whose writing systems conflict with high-resource pretraining priors. We study Bai-Chinese translation in the Bowen script, where Chinese characters are frequently borrowed phonetically rather than semantically, causing severe literal-translation hallucinations. We propose MAS-Bai, a linguistically informed adaptive multi-agent framework for Bowen-script Bai-Chinese translation. MAS-Bai routes each input through either a lightweight retrieval-translation path or a deeper philological path with etymological and syntactic agents, and uses cycle-consistency with a lightweight reverse translator as an inference-time reliability signal. We construct a 6,536-sentence Bai-Chinese corpus and evaluate on a 1,536-sentence test set. MAS-Bai achieves 16.23 BLEU and 62.15 COMET, improving over Standard RAG by 8.11 BLEU and 6.85 COMET. Forward LoRA, expert ablations and controlled routing experiments distinguish the roles of supervised adaptation, Bai-specific expertise and conditional routing. The largest gains occur on Bowen phonetic loans and divergent Bai word order, while human evaluation and reverse-verifier analyses support the automatic results.

0.35

Measurement-Adaptive Light-Cone Transform for Robust Non-Line-of-Sight Imaging

eess.IV cs.CV ·2026-10-16 ·Lianfang Wang, Xueying Liu, Junnan Dong, Qinghua Zhang, Jun Liu +2 more
published
Abstract

Practical non-line-of-sight (NLOS) measurements are affected by noise and deviations from idealized light transport, which can degrade reconstruction with conventional light-cone transform (LCT) methods. We propose the Measurement-Adaptive Light-Cone Transform (MALCT), a physics-guided framework that retains the efficient Fourier-domain inversion structure of LCT. {Within a unified variational formulation, MALCT combines the analytical response with a shared learned response through measurement-conditioned weighting and predicts a measurement-conditioned, frequency-dependent regularization map.} An attention-enhanced spatio-temporal feature extraction module captures long-range temporal dependencies and local spatial correlations to support reconstruction under low-signal conditions. Experiments on synthetic data and real measurements from different NLOS systems demonstrate improved reconstruction quality. Controlled evaluations further assess robustness to temporal-response broadening and reduced acquisition time. MALCT also preserves scene structures with spatial sampling grids down to $32\times32$, supporting its potential for NLOS imaging under limited acquisition budgets.

0.35

ANCHOR: Audio-Visually Grounded Chain-of-Thought Reasoning Benchmark

cs.CV ·2026-10-16 ·Joel Julin, Souraja Kundu, Liza Dahiya, George Z. Wei, Liuyue Xie +4 more
top venuebenchmark
Abstract · Accepted at NeurIPS 2026, Evaluations & Datasets Track. Joel Julin and Souraja Kundu are co-first authors

Multimodal large language models (MLLMs) perform well on audio-visual question answering, but correct answers do not necessarily imply that their stated reasoning is grounded in relevant perceptual evidence. Existing benchmarks mainly evaluate final-answer accuracy, offering limited insight into whether model-generated rationales correctly identify visual regions, temporal intervals, and audio events supporting predictions. We introduce ANCHOR, a benchmark and evaluation framework for measuring grounded-rationale alignment in multimodal reasoning, comprising 2,319 videos and 8,272 question-answer pairs with structured reasoning traces, temporal evidence, spatial bounding boxes, and salient audio cues, including a human-refined test set of 224 videos and 975 question-answer pairs. We separately measure answer correctness, hallucination, logical divergence, and conciseness, with direct spatial and temporal grounding measurements on a focused subset. Across the full benchmark, twelve contemporary MLLMs show a substantial alignment gap: models with comparable answer performance can differ markedly in whether their explanations are supported by annotated evidence. On a matched subset of 100 videos and 446 question-answer pairs, rankings remain highly stable across human-refined, Gemini-generated, and GPT-generated references (mean Pearson r=0.99), providing evidence against model-reference circularity within this subset. Grounding-cue ablations of three MLLMs on the full human-refined test set reveal a consistent modality hierarchy: removing video reduces accuracy by 38.4-41.8 points, while removing bounding boxes (3.2-7.7 points) and audio (4.2-5.6 points) causes smaller, complementary, model-dependent drops. These findings show answer-only evaluation obscures important grounding failures and position ANCHOR as a testbed for evaluating accurate and evidence-aligned audio-visual reasoning.

0.35

On the Tip of the Tongue: First-Token Cueing as a Behavioral Probe of Recall Failures in LLMs

cs.CL ·2026-10-16 ·Jaewon Sok, Jewon Yeom, Seonghyeon Park, Jeongjae Park, Hwiyeong Lee +1 more
benchmark
Abstract

Large language models often answer factual questions incorrectly in closed-book settings, and such errors are usually attributed to missing knowledge. Yet models often encode answers they fail to produce. We show that many closed-book errors are first-token recall failures: the model commits to a wrong first token, but once given the right one it completes the answer and often recognizes it as correct. We study this with First-Token Cueing, a behavioral probe that prefills the response with the first token of the gold answer. Model uncertainty peaks at the first answer token, where it already predicts correctness, and among gold prefixes of any length the first token alone accounts for the largest share of recovered errors. Across 16 open models, this one token corrects between a fifth and a half of the errors on multi-token answers in English QA and improves accuracy in all seven languages of a multilingual benchmark. Most of the errors it corrects are also solved by the model's own sampled reasoning, a criterion for encoded knowledge, and models recognize correct starts they do not produce: they continue a correct start more confidently than an incorrect start of the same answer type and prior probability, yet rarely find it without the gold answer. These results locate a recoverable class of factual errors at the first decoding step, where inference-time methods may gain the most.

0.35

BRICE: Brain-inspired Recurrent Illusory Contour Encoder for Point Cloud Completion

cs.CV ·2026-10-16 ·Yunlei Wang, Dawei Li, Xue-song Tang
code/data
Abstract

Mainstream point cloud completion methods are unanimously based on deep learning architectures. On one hand, these methods tend to repetitively stacking attention modules to enhance network performance, which drifts away from the neat design of unified and interpret-able networks; on the other hand, nearly all existing architectures rely solely on feed-forward connections to propagate information, neglecting feedback from deeper layers. The recurrent neural architecture, formed by feedforward-feedback loops, have shown its ability to generate Illusory Contours (ICs) in the V1 brain region of mammals; and ICs are essentially a completion mechanism of the brain for visual information. Motivated by this, we design a Brain-inspired Recurrent Illusory Contour Encoder (BRICE) to "imagine" what is missing in the incomplete 3D point clouds. Building upon the foundation model of Predictive Coding (PC), the plug-and-play nature enables BRICE to be seamlessly integrated into mainstream point cloud completion baselines. Experimental results show that BRICE can enhance various completion baselines on different 3D datasets such as MVP2K, PCN, and PlantPCom. Our code: Https://github.com/wangyunlei-wyl/BRICE.

0.35

Image Moment Estimation for Crowded Datasets

eess.IV q-bio.QM ·2026-10-16 ·Andy Zhang, Oscar Mickelin, Amit Singer
Abstract · 29 pages, 11 figures

Single-particle cryogenic electron microscopy (cryo-EM) is an imaging technique capable of reconstructing high-resolution three-dimensional (3D) structure of biological macromolecules from a set of noisy projection images taken at random viewing directions. The mathematical model for cryo-EM images typically assumes that each image contains exactly one particle, which is centered at the origin. However, many experimental datasets contain particle images which include multiple particles: one central particle and a few crowding particles. This deviation from the model leads to inaccurate estimation of image statistics, such as the low-order image moments. To account for this, we propose a method to estimate image moments while correcting for the effect of crowding particles. This method can be run directly from a set of cryo-EM micrographs with picked particles, and we demonstrate its efficacy on a wide array of synthetic data.

0.35

ConvexLLM: Scaling Convex-Hull Speech Interfaces to General-Purpose Audio Intelligence

cs.CL cs.SDeess.AS ·2026-10-16 ·Ming-Hao Hsu, Shujie Liu, Yuxuan Hu, Yuxiang Wang, Yingda Shen +2 more
Abstract

Speech and audio language models connect continuous acoustic representations to decoders pretrained primarily on text data. Most such models use MLP or Q-Former interfaces to map acoustic features into the decoder's input space, requiring the decoder to learn how to interpret the resulting representations. We investigate whether grounding these representations in the decoder's existing text embeddings can help it use acoustic information while better preserving its pretrained text capabilities. We present ConvexLLM, a general-purpose speech and audio language model whose C-Gate interface represents audio as continuous convex combinations of frozen text embeddings. ConvexLLM achieves an eleven-task VoiceBench aggregate of 75.36, the highest among the open-weight models we compare, and an MMAU Test-mini accuracy of 62.50%. Its largest margins over the baselines are on spoken knowledge and reasoning tasks, while the strongest baselines are more accurate on audio understanding and transcription. In a comparison with a 2B-parameter decoder, C-Gate achieves higher aggregate spoken-language performance than MLP and Q-Former interfaces, with text-task performance close to MLP and task-dependent tradeoffs in transcription and audio understanding. These results support convex combinations of text embeddings as a viable interface for general-purpose speech and audio language models.

0.34

Competence-Based Evaluation of Logical Reasoning in Large Language Models

cs.CL cs.AIcs.LG ·2026-10-16 ·Jize Jiang, Chengxiang Zhai
position
Abstract

Recent advances in large language models (LLMs) have brought impressive performance across diverse tasks; however, shortcomings remain in logical reasoning, including inconsistencies under linguistic variation. To improve the reasoning capacity of LLMs, we must be able to accurately measure their reasoning capacity. Current benchmarks primarily report single accuracy numbers on selected reasoning tasks. Such a performance-based evaluation methodology says little about a model's generalization capacity in reasoning. To address this limitation, we argue for a different evaluation paradigm, called competence-based evaluation (CBE), which tests whether models reliably apply the same competence across a family of problems. A parameterized generator scales complexity while keeping the required competence fixed, and matched answer-preserving presentations test correctness beyond one formulation. We instantiate CBE on elementary ordering tasks, scaling input size while varying presentations of the same ordering rule. Across six GPT and Qwen models, performance is reasonably good over bounded complexity ranges, but even strong models fail the competence condition across the tested family. Chain-of-thought and GRPO extend strong performance without resolving all tested failures; supervised LoRA shows little held-out transfer. CBE provides diagnostic profiles and an evaluation protocol that can grow with future models.

0.34

NV-Generate-Body: 3D Latent Diffusion Model forControllable Anatomy, Tumor, and Demographic Body Mask Synthesis

cs.CV ·2026-10-16 ·Can Zhao, Pengfei Guo, Dong Yang, Yufan He, Yucheng Tang +6 more
code/data
Abstract · Code and model weights open-sourced

Curating large labeled CT datasets is expensive and constrained by patient privacy. We present NV-Generate-BodyMask, a 3D latent diffusion model that synthesizes whole-body CT segmentation masks with precise, joint control over organ anatomy, tumor burden, and patient demographics. The model operates in the compressed latent space of a frozen mask autoencoder, generating $256^3$ volumetric label maps conditioned on a 19-dimensional vector via Adaptive Group Normalization and cross-attention. A size-adaptive ROI loss with spatial halo weighting and a paired tumor injection pipeline enable accurate tumor size conditioning despite the scarcity of annotated tumor cases. Trained on 8,571 masks across 26 cohorts spanning 132 labels, the model achieves Spearman $ρ= 0.70$--$0.92$ for organ size, 85--100\% tumor recall for lung, pancreatic, colon, and bone lesions, and strong demographic alignment (BMI $ρ= 0.78$, sex-conditioned perfect prostate control).

0.34

BanglaShield: A Script-Robust Safety Classifier and Benchmark for Bangla Native, Romanized, and Code-Mixed Prompts

cs.CL cs.LG ·2026-10-16 ·Mohsinul Kabir, Md. Tariquzzaman, Mohammad Rashidujjaman Rifat, Tharindu Ranasinghe, Nishat Raihan
benchmarkcode/datahot direction
Abstract · 17 pages. Code and data: https://github.com/mraihan-gmu/BanglaShield-Model-Benchmark ; dataset: https://huggingface.co/datasets/md-nishat-00…

Safety classifiers for large language models (LLMs) are typically built and evaluated on standard-script text, yet Bangla, like many lower-resource languages, is routinely written in native script, in Latin-script romanization, and code-mixed with English, Hindi, or Urdu. We introduce BanglaShield, a resource and model for graded prompt-safety classification across these written forms. We consolidate 10,245 Bangla prompts from six sources under a four-level taxonomy (Safe, Offensive, Harmful, Severe) and render each prompt in three aligned forms that share labels and a group-aware split, enabling strictly paired comparisons. Benchmarking six contemporary LLMs, including a Bangla-specialized one, on 36,882 classification queries and 24,318 judged responses, we find that LLM classifiers label 47.0% of unsafe native-script prompts as Safe, rising to 63.4% under romanization, while response-level safety accuracy drops by 10.3 points. Three multilingual guard models also pass 30.3-57.3% of unsafe native-script prompts. Using a model's own classification as an input filter reduces harmful compliance by only about five points. In contrast, BanglaShield, a compact fine-tuned encoder, reaches 87.8 macro-F1 on native-script and 84.4 on romanized prompts, outperforming the best LLM on each form by 42.8-46.1 points with ~130x fewer parameters than Gemma 4 31B. We release the corpus, splits, judgments, and model.

0.34

CatchUp: Depth-Aware Re-Admission of Visual Tokens for Efficient Large Vision-Language Models

cs.CV ·2026-10-16 ·Eunju Park, Uiwon Hwang
publishedcode/data
Abstract · WACV 2027. Code: https://github.com/pej0918/CatchUp

Large Vision-Language Models (LVLMs) repeatedly process hundreds of visual tokens across decoder layers, incurring substantial computation and memory costs. Existing visual-token reduction methods often irreversibly discard tokens despite depth-varying query relevance. Although recoverable routing allows excluded tokens to return, re-admitted tokens retain stale hidden states after bypassing intermediate decoder transformations. Thus, token recoverability does not guarantee representation readiness at the current decoder depth. Under a fixed routing trajectory at a 64-token budget, replacing stale re-admitted states with current-depth oracle representations improves average RefCOCO Acc@0.5 from 18.96 to 23.18, revealing the impact of representation staleness. We propose CatchUp, a training-free framework combining reversible visual-token routing with Local Transition Correction (LTC). CatchUp repeatedly re-evaluates query-conditioned relevance over the complete visual-token set, preserving deferred tokens for later re-admission. LTC approximates their skipped representation evolution by transferring reference-to-current hidden-state transitions from semantically and spatially related active anchors. This correction improves alignment with current-depth oracle representations without recomputing the skipped decoder layers. Across three referring-expression grounding datasets and six multimodal benchmarks, CatchUp achieves the best grounding performance across all eight splits and the highest average multimodal retention among evaluated efficient methods at 64-, 128-, and 192-token budgets. At a 64-token budget, CatchUp processes 88.9% fewer active visual tokens per post-routing decoder interval, retains 92.17% of dense-model multimodal performance, and reduces end-to-end latency by 8.0%. Code is available at https://github.com/pej0918/CatchUp

0.34

COMET: Decoupled Distillation, Routing, and Capacity Control for Task-Agnostic Continual Vision--Language Learning

cs.CV cs.AI ·2026-10-16 ·Seungyoon Woo, Keonhee Park, Gunhee Kim
top venuebenchmarkposition
Abstract · Accepted at NeurIPS 2026

Vision-language models (VLMs) such as CLIP retain zero-shot recognition ability, but their open-world use often requires task-agnostic continual learning: tasks arrive sequentially, past data are unavailable, task identity is unknown, and the backbone cannot be retrained. We argue that this setting fails VLMs through three coupled pressures beyond the stability-plasticity tradeoff: (1) modal-consistency drift, (2) representation-space interference, and (3) per-task capacity shortfall. We propose COMET (Collaborative Mixture-of-Expert Transfer) to resolve these three issues respectively as follows. A frozen teacher provides only a feature reference on a shared image-text pool; Tri-Affinity Distillation preserves image-image, text-text, and image-text geometry; and Student-Aware Cross-Expert Distillation aligns new experts with compatible prior experts so routing errors degrade gracefully. We design COMET to keep capacity and inference separate: a contextual bandit keeps or merges full per-task experts using post-training signals, while a reconstruction router selects sparse experts at test time without teacher semantics. On the X-TAIL benchmark, we show that each proposed component improves continual learning performance. COMET shows that continual vision-language adaptation can preserve multimodal geometry, coordinate accumulated experts, and allocate capacity jointly when feature reference, expert shaping, routing, and capacity control are separated by construction.

0.34

Beckmann World Models: Direct Terminal-Map Learning for One-Step Action-Conditioned Prediction

cs.CV ·2026-10-16 ·Adam Lee, Shobhit Agarwal, June Pyo Suh
Abstract

Action-conditioned world models support planning and policy evaluation by predicting future states. However, learning an accurate one-step predictor can require substantial training computation. We introduce Beckmann World Models (BWM), which learn a direct map from noise, visual history, and actions to future frames without a pretrained generative teacher. We adapt the terminal-map conservation law of Beckmann Transport Models to action-conditioned video prediction, training directly from noise-data pairs without conditional populations of generated samples required by DriftWorld. However, transport consistency on interpolated inputs does not ensure accurate prediction from the pure noise used at inference. We address this gap with endpoint-aligned training that directly supervises the source prediction, emphasizes motion and spatial structure, and teaches recovery from generated histories. All components train the same map and retain one network call per future chunk. On PushT and Robomimic Can, the final model reduces 64-frame prediction MSE by 68% and 57%, respectively, relative to same-backbone regression baseline. For policy evaluation, BWM achieves the lowest mean IoU error among the evaluated models, 0.0100, at 24.8 ms per four-frame PushT prediction. On Robomimic Can, the BWM stage trained with motion supervision and generated histories achieves 3.3 times higher throughput and 78% lower peak memory than local one-step DriftWorld. In a latent-space extension to RT-1 and Bridge-V2, equal-update controls show that feature supervision under generated histories improves perceptual prediction. These findings show how terminal-map learning can combine efficient training with accurate one-step visual prediction.

0.34

SERO: Semantic Prediction for Single-Step Object Removal

cs.CV ·2026-10-16 ·Amreen Hasheem Siraj, Alireza Javanmardi, Pragati Jaiswal, Didier Stricker, Alain Pagani
publishedcode/data
Abstract · Accepted at CVMP 2026. Project page: https://dfki-av.github.io/SERO

Object removal and semantic inpainting of missing regions remain inherently ill-posed, as multiple plausible completions may exist for a given scene. Recent diffusion-based methods have demonstrated strong potential for high-quality object removal; however, they still face two key limitations. First, although they can synthesise sharp and high-resolution images, they often struggle to complete removed regions in a semantically consistent and contextually meaningful manner, where the generated content should agree with the surrounding scene layout and object relationships rather than merely appear visually realistic. Second, their iterative denoising process makes them computationally expensive, limiting their practicality for downstream applications. We propose SERO, a semantic prediction framework for single-step object removal. Our model predicts a hidden-region semantic prior in embedding space, which guides a latent bridge-matching generator toward plausible scene completion by providing both where to edit and what structure is likely to exist behind the removed object. This enables visually realistic and contextually coherent completions in one forward pass. Experiments on the RORD and OBER datasets show that SERO achieves competitive or state-of-the-art performance while requiring only a single generation step.

0.33

SynerGest: Temporal-Spatial Synergistic Gesture Generation for Dyadic Conversations

cs.CV cs.SDeess.AS ·2026-10-16 ·Chengxiang Guo, Jian Wang, Lanmiao Liu, Zhan Tang, Junhua Fei +1 more
top venue
Abstract · accepted by ACM MM 2026

Co-speech gesture generation in dyadic conversations for creating immersive virtual avatars remains highly challenging. Existing models often struggle to balance kinematic stability with rhythmic responsiveness and suffer from semantic entanglement, leading to either generic mean-pose regression or unnatural micro-jitters. To address these challenges, we propose SynerGest, a high-fidelity dyadic gesture generation framework grounded in Contrastive Flow Matching (CFM). Specifically, our CFM objective explicitly penalizes mismatched semantic pairs during training, discouraging spatial shortcuts and over-smoothed predictions while improving conditional semantic alignment. Furthermore, to overcome architectural bottlenecks, we introduce the Impulse-Trend Synergistic Attention (ITSA) module. Grounded in the key insight of decoupling "what moves" from "when it moves," ITSA models sharp, speech-driven rhythmic impulses separately from continuous kinematic trends, supporting precise audio-motion synchronization and global structural stability. Extensive qualitative and quantitative evaluations on two public datasets demonstrate that SynerGest achieves state-of-the-art performance, delivering contextually responsive and kinematically fluid dyadic interactions.

0.33

FlexPart: Text-Controlled Semantics-Aware 3D Part Segmentation at Flexible Granularity

cs.CV ·2026-10-16 ·Yiqun Zhao, Jiachen Xu, Yihao Zhi, Yunfei Zhao, Zibo Zhao +3 more
Abstract · 8 pages main, 8 figures

A 3D asset admits multiple valid part decompositions, while semantics-aware segmentation should reflect both the desired granularity and the parts a user wants to extract. Ideally, a practical segmentation model should strive for a simple principle: what you ask for is what you get. However, realizing this goal is hindered by two core obstacles: the lack of multi-granularity training data and the absence of a holistic control mechanism. We propose FLEXPART, a unified framework for text-controlled 3D mesh segmentation. To establish flexible control, we introduce a holistic control prompt comprising semantic part names with location and shape clauses. To support this mechanism, we utilize an automated Vision-Language Model (VLM) pipeline to construct a large-scale dataset of multi-granularity partitions and their corresponding prompts. Driven by this data, FLEXPART generates a view-independent part-identity field in native 3D space, directly assigning discrete part IDs to mesh faces. To compel the model to respect semantic conditions rather than rely on geometric priors, we train the model using randomized background dropping. Furthermore, an inference-time VLM planner can automatically complete descriptive clauses from 2D renderings, allowing users to simply name the desired parts in ordinary words. FLEXPART with planner-completed prompts outperforms the strongest evaluated baseline by 15% in mIoU, even when that baseline receives ground-truth guidance. On our held-out FLEXPARTBENCH, FLEXPART leads at each granularity and improves average mIoU by 22% over the strongest baseline. With only the requested part names supplied by the user, it also improves IoU by 35% over the strongest click-driven baseline on selective segmentation.

0.33

LSNR: Learned Sparse News Recommendations with Attentive User Modeling

cs.IR ·2026-10-16 ·David Vos, Andrew Yates, Maarten de Rijke
top venue
Abstract · 10 pages, 3 figures, 9 tables. Accepted at SIGIR-AP 2026

Current learned sparse approaches for recommender systems, such as STAR, represent a user by concatenating their interaction history into a single query string. For news, this creates two bottlenecks: concatenation collapses a user's distinct reading interests into one flat query, and, because news articles are long-form text, concatenating several news articles quickly overruns the context window of the language model used as backbone and forces quality-degrading truncation. To address both bottlenecks, we introduce learned sparse news recommendations (LSNR). LSNR encodes each article independently and constructs a sparse user vector by combining learned, attention-based weighting of the per-article sparse representations with a dense-projected expansion at the user level. Experiments on the MIND and EBNeRD datasets show that LSNR significantly outperforms STAR, with these gains holding even in short-history settings where truncation is not a factor. Compared to state-of-the-art dense recommenders, LSNR is competitive in standard settings and it outperforms them when the user history is short. An ablation of the user encoder confirms that learned attention-based aggregation outperforms simpler pooling alternatives. Beyond effectiveness, an additional advantage of LSNR is that it serves from an index an order of magnitude smaller than an uncompressed dense baseline.

0.32

Bengali Hallucination Detection with Retrieval and Programs: A System Study

cs.CL ·2026-10-16 ·Shamim Hasan, Samiya Ahmed Chowdhury
benchmark
Abstract · 8 pages

Bengali hallucination detection must distinguish unsupported factual claims from incorrect reasoning, often without a supplied reference passage. We present and characterize an inference-only system that combines response-conditioned query rewriting, hybrid retrieval, and task-specific language-model judging. Six retrieval routes search question-answer and passage collections; a router directs each example to a 27B textual judge or a 14B agent that can compare retrieved answers, reason directly, or generate and execute Python. We characterize the system using version-specific code, saved predictions, execution traces, and corpus metadata from the Olikbochon challenge. Kaggle associates a displayed private macro-F1 of 0.956 with Version 2; its saved run processes 2,516 examples in 220.99 minutes on two T4 GPUs. Of 253 math-routed examples, 88 terminate through candidate matching, while execution traces expose 49 stderr-bearing run events across 35 examples. A later run changes six predictions despite identical core worker code. A bounded audit of 1,090 benchmark-related corpus records finds exact question or context overlap for 38 processed evaluation inputs, precluding an unqualified claim of corpus-independent generalization. The contribution is a concrete retrieval-and-computation judging architecture with an auditable operational characterization, not a controlled estimate of component gains. We distinguish the reported leaderboard result from independently recomputed diagnostics and identify the evidence needed for stronger effectiveness and reproducibility claims.

0.32

Ignored, Not Refused: What Low Prompt-Injection Success Rates Hide in Small Tool-Using Agents

cs.CL cs.CR ·2026-10-16 ·Farhan Zarif, Kazi Shahid Ahmed Galib, Synthia Akter Promi, Shabnam Sharmin Tarin, Ariyan Hossain +2 more
Abstract · 17 pages, 3 figures, 11 tables

Tool-using agents read untrusted external content, which exposes them to indirect prompt injection (IPI), where hidden instructions hijack the agent's tool calls. Benchmarks measure IPI security by the attack success rate (ASR), which treats every failed attack as a successful defense. To test this assumption, we audit 33,000 trajectories from eleven open-weight agents (2-14B parameters) on InjecAgent and AgentDojo. A label-blind LLM judge shows that most failed attacks are ignored (76-80%) rather than refused (17-18%). With identical model weights, changing only the system prompt shifts InjecAgent ASR from 0.2% to 43%. Our Attack-Task Outcome Matrix (ATOM) further shows that in 21.7% of AgentDojo episodes, the agent stayed safe but did not complete the user's task. A low ASR therefore mostly reflects agents silently ignoring injections, not refusing them. Evaluations should report how attacks fail and whether the user's task is completed, alongside ASR. We release the labeled corpus, judge prompt, and analysis code.

0.32

DICL-Qwen3VL: Decoupled Confidence Calibration for Interpretable Workwear Detection

cs.CV ·2026-10-16 ·Meng Wang, Wu Zeng, Mei Li, Chao He
code/data
Abstract

Poor illumination and occlusion obscure clothing cues in indoor work sites, while diverse clothing styles complicate workwear compliance judgments. Although vision-language models (VLMs) support localization and semantic reasoning, their predictions often lack explicit supporting evidence, and a single confidence score cannot distinguish visual uncertainty from ambiguity in compliance judgments. We propose DICL-Qwen3VL, a decoupled interpretable VLM that integrates structured workwear detection, evidence generation, and separate visual and reasoning confidence estimates. First, low-rank adaptation (LoRA) adapts Qwen3-VL to localize visible upper-body clothing and identify workwear. Second, decoupled calibration separates visual observations from compliance reasoning and assigns a confidence score to each. Token entropy and sensitivity to image perturbations provide supervision for visual confidence, linking confidence estimation to the generated observations and their dependence on image content. Finally, group relative policy optimization (GRPO) jointly optimizes detection accuracy, output validity, and confidence calibration. Experiments on the company-collected Indoor Construction Site Workwear Dataset (ICSWD) and the web-collected Indoor Construction Workwear Dataset (WICWD) yield mAP@50 values of 0.730 and 0.753, respectively, outperforming the evaluated LoRA-adapted VLM baselines, including Qwen3-VL. The framework also provides explicit evidence and separate confidence scores to support manual review in industrial safety monitoring. WICWD is publicly available at https://github.com/DaErwen1/WICWD.

0.32

Agentic Physics-Grounded Video Generation via Memory-Augmented In-Context Reasoning

cs.CV ·2026-10-16 ·Xikang Guan, Shaocong Yang, Yanqi Bao, Can Wang, Zheng Gu +4 more
code/data
Abstract · 12 pages, 5 figures, 1 table

Physical plausibility is essential for realistic video generation. Existing methods impose physical guidance before synthesis via simulations or refined prompts, but this can limit dynamic diversity, miss errors that emerge only after generation, and require repeated regeneration when inconsistencies remain. We therefore study a generate-diagnose-repair paradigm that preserves the original generation capability and selectively repairs physical inconsistencies afterward, raising two key questions: what to repair and how to repair? To address them, we introduce PhysRepair, an agentic video generation framework that combines causal reasoning over the observed physical failure with prior repair experience to enable localized and adaptive correction. Specifically, Spatiotemporal Causal Constraints identify the first invalid physical state and its causally affected states, producing a repair specification that guides physically plausible local regeneration while preserving unaffected content. Experience-augmented In-context Learning combines the ongoing repair context with relevant experience from previous repairs, allowing the agent to adapt its repair decisions across diverse physical failures and accumulate completed trajectories for future cases. Experiments on Physics-IQ Verified across solid mechanics, fluid dynamics, optics, thermodynamics, and magnetism demonstrate that PhysRepair consistently improves the physical plausibility of generated videos over existing methods. Our project page is available at https://xxkkxxx.github.io/PhysRepair-Website/.

0.32

UniField++: A Unified Agentic Framework with Loop Reasoning for MRI Field-Strength Enhancement

cs.CV ·2026-10-16 ·Yiyang Lin, Chenhui Wang, Tianliang Yao, Pengli Zhu, Shuaibo Li +3 more
code/data
Abstract · 10 pages, 6 figures, 8 tables

MRI field-strength enhancement seeks to combine the accessibility of low-field scanners with the image quality of high-field scanners. Yet, existing methods remain prone to anatomical hallucinations, as uniform processing leaves difficult brain regions insufficiently examined and scarce paired data limit their learning of how tissue contrast and anatomical visibility change across field strengths. To address these challenges, we propose UniField++, a two-agent framework that unifies enhancement tasks across multiple field-strength transitions, including 64mT-to-3T and 3T-to-7T. Unified learning allows both the VLM-based planner and the flow-matching-based enhancer to draw on field-dependent image characteristics observed across tasks, rather than relying on a single transition. Trained with radiologists' annotations of hallucination-prone regions and corresponding correction plans, the planner predicts regional risk and formulates targeted restoration strategies. Guided by these plans, the enhancer performs loop reasoning along the flow-matching trajectory, repeatedly processing high-risk regions for additional refinement. Together, cross-transition learning and risk-guided refinement support anatomically faithful enhancement across field strengths. Across the evaluated tasks, UniField++ improves PSNR by an average of 2.74 dB and SSIM by 14.05 percentage points over comparison methods, with corresponding gains of 3.97 dB and 13.92 percentage points over the ablation baseline. It also maintains enhancement quality on an external cohort with an unseen field strength and improves downstream whole-brain segmentation Macro Dice by an average of 0.1457 over the original low-field images. Code is available at https://github.com/CUHK-AIM-Group/UniField_plusplus.

0.31

I2V-DETACH: Source Grounding Detachment for Unauthorized Image-to-Video Generation

cs.CV ·2026-10-16 ·Chanhui Lee, Yeonghwan Song, Yewon Kang, Jeany Son
published
Abstract · NeurIPS 2026

Recent diffusion models have advanced image-conditioned video generation, but their accessibility raises risks of unauthorized manipulation. Image immunization has emerged as a promising defense for Image-to-Video (I2V) protection, yet existing methods often rely on degrading visual quality or temporal coherence. In this work, we revisit I2V immunization as reducing source-content preservation, shifting the focus from output degradation to weakening source-conditioned content propagation. To this end, we propose I2V-DETACH, an image immunization method that prevents unauthorized I2V generation through source grounding detachment. I2V models preserve source content by propagating source-image cues into video latents during denoising. Accordingly, I2V-DETACH disrupts this process in two complementary ways: it suppresses source-video grounding to reduce source cue propagation, and strengthens intra-video interactions to shift denoising toward video-side dynamics with reduced reliance on the source image. To obtain informative optimization signals, we initialize auxiliary video latents with source-oriented states that remain within the source-conditioned generation regime while deviating from source-preserving trajectories. Extensive experiments across diverse diffusion architectures show that I2V-DETACH substantially reduces source-content preservation, improves over prior immunization methods, and exhibits strong black-box transferability.

0.31

AgentCount: An Agentic Workflow for Traceable Colony Counting and Insight Generation from Agar Plate Images

cs.CV q-bio.QM ·2026-10-16 ·Zhuosheng Liu, Austin E. Weber
code/dataunusual
Abstract

Colony counting is commonly used to evaluate microbial populations in food safety testing and microbiological studies. Despite its wide application, manual counting is time-consuming and operator-dependent, and the final count often lacks a record of the objects included in the measurement. To address these limitations, this study presents AgentCount, an open-source agentic workflow for colony counting and result interpretation from agar plate images. Plate photographs and plating metadata are processed through seven agents for image quality control, plate localisation, colony detection, counting, CFU calculation, validation, and reporting. Colony detection is performed using Segment Anything Model 3 (SAM 3) with multiple text prompts and morphology-based filtering. After that, replicate statistics are calculated and a large language model (LLM)-based agent is used to generate a results narrative. Intermediate outputs, including quality scores, segmentation masks, detections, and warnings, are retained to support review of the reported values. The workflow also extends to skill-guided analysis of archived microbial population data. In a retrospective comparison with two previously published studies, we benchmarked 241 published endpoints, including all 216 mean +- SD values of the meat-analogue pathogen tables; on the 168 endpoints with a complete n=6 replicate source, 164 (97.6%) agreed within 0.01 log10 units. Overall, AgentCount demonstrates the promising potential of agentic workflows for applications in applied microbiology.

0.31

Implicit Neural Models for Direct Multi-Shell Diffusion MRI Signal Estimation

eess.IV cs.CV ·2026-10-16 ·Sanna Persson, Fabian Leander Sinzinger, Christoffer Olsson, Rodrigo Moreno
Abstract · 23 pages (9 pages main text + supplementary material), 9 figures, 18 tables

Diffusion magnetic resonance imaging is a noninvasive technique that is sensitive to the motion of water molecules in human tissue. The image signal is commonly described in q-space; however, a complete coverage of this space is often infeasible due to hardware and acquisition time limitations. In this work, we propose to represent the entire diffusion signal using an implicit neural representation network with wavelet activation functions. The signal space is encoded using Fourier features and spherical harmonics basis functions for a continuous output distribution of the raw diffusion data, enabling both super-resolution and q-space interpolation. Additionally, we introduce a pretraining strategy using a group template, enabling faster per-subject optimisation when combined with nonlinear registrations of subjects into template space. We tested the method with 10 subjects from the Human Connectome Project and 26 subjects from in-house data (multidimensional diffusion MRI from a Parkinson's disease cohort). The results show that the proposed method yields high-fidelity reconstructions with significantly reduced memory requirements compared to raw diffusion-weighted images. The pretraining scheme showed good generalisability properties on data acquired with very different protocols. We further assess the method on downstream metrics relevant to diffusion MRI data from the learned representations.

0.31

The Reliance Hypothesis: Chain-of-Thought Faithfulness Rises as Language Resource Level Falls

cs.CL cs.AI ·2026-10-16 ·Lucas Resck, Jingyi Sun, Qianli Wang, Isabelle Augenstein, Anna Korhonen
Abstract · 21 pages, 10 figures, 16 tables

Chain-of-Thought (CoT) faithfulness matters for monitoring, explaining and safely deploying large language models, yet it has been studied mostly in English. Languages differ in level, the amount of text available to learn from, so models know less in lower-resource languages, which may change how much they rely on their reasoning. To our knowledge, we are the first to relate CoT faithfulness to both resource level and task type (knowledge- or reasoning-intensive). On parallel multiple-choice questions from Multilingual Massive Multitask Language Understanding (MMMLU) in eight languages, faithfulness rises as resource level falls on knowledge-intensive tasks but not on reasoning-intensive ones, a gap that points the same way across four models and several faithfulness measures. We propose, and find support for, the Reliance Hypothesis: models rely more on their reasoning when they cannot recall the answer, as in lower-resource languages on knowledge-intensive tasks. Model confidence partly explains the effect of resource level, but not the knowledge-reasoning gap. Languages in which models are less confident rely more on reasoning, yet less confident answers within a language do not, a Simpson's paradox. In lower-resource languages, models rely on their reasoning for more answers, not more strongly for each. CoT monitoring may therefore be most informative in lower-resource languages, where it is least often validated.

0.31

Do Speech BCIs Need Larger Models? Rethinking Neural Decoding beyond Scaling

cs.CL ·2026-10-16 ·Zehui Feng, Weichuan Wang, Xiaohan Chen, Cuntai Guan, Ting Han
top venue
Abstract · 25 pages, Accepted by NeurIPS 2026 as a Main Track conference paper

Recent speech brain-computer interfaces (BCIs) increasingly rely on large models and complex multi-stage pipelines for neural decoding. However, in speech BCI, where training data are scarce, noisy, and expensive to collect, the benefits of model scaling remain unclear. This work revisits a fundamental question: are large models truly necessary for accurate neural decoding? We propose BEST (Brain-Encoding-Speech-Text), a two-stage framework for efficient neural-to-text decoding. Instead of relying on large generative models or phoneme-centric pipelines, BEST introduces hierarchical intermediate representations and a compact speech-oriented decoder that directly maps neural activity to text. This design preserves essential structure while significantly reducing computational cost, enabling robust generalization in low-resource settings and practical deployment under constrained resources. On the Brain-to-Text '24 and '25 benchmarks, BEST achieves word error rates (WERs) of 6.19% and 1.62%, respectively, using only 3.85% of the parameters required by comparable methods. Beyond these results, we provide three key insights: (1) decoder scaling yields non-monotonic gains; (2) cross-modal alignment is more critical than model size; and (3) overfitting degards sequence-level performance. Representation analysis further reveals a consistent neural-audio-text alignment pathway.

0.31

MuninAI: A Self-Hosted Agentic Framework for Independent Research Groups

cs.CL ·2026-10-16 ·Markus Fischer, Jens Meiler, Matthias Elgeti
code/data
Abstract · 37 pages, 3 figures, 20 tables. Code, evaluation suite and scorecards: https://github.com/elgetilab/munin (archived release v1.0.0: https://…

Research groups that want a language model grounded in their own literature face a choice between hosted services that receive their unpublished work and self-hosted deployments that keep the data local but give the model no access to the corpus. Munin is an open-source framework for a self-hosted scientific assistant that indexes a group's corpus and its citation graph, and wraps whichever open-weight model the group can serve in an agent harness with specialised reading, search, and computation tools. We claim that the harness, rather than the specific model, is what makes the system useful on domain literature, and we ship the evaluation suite that tests this claim with the software. On 199 paired literature questions, the harness more than doubles the accuracy of the base model, from 0.387 to 0.874 (+0.487, 95% CI [0.407, 0.568]), while naive retrieve-then-answer scores below the unaided model on every backbone tested. Under the harness, roughly twice as many answer claims trace to a passage the model read itself as with fixed top-5 retrieval. When the source paper for a question is removed from the corpus, the system abstains on about nine in ten of the questions it had answered correctly on the Qwen backbones and eight in ten on gpt-oss-20b, and it produces no fabricated local citation across 300 fabricated references. The pattern holds on three backbones from two developers, including a 20-billion-parameter model that runs on a single consumer-grade GPU. With all outbound network access disabled, both Qwen backbones still exceed the published PaperQA2 accuracy. The evaluation suite runs as a single command, so an adopting group can re-run it on its own corpus and model.

0.31

PREEMPT: Proactive Robustness Evaluation under Editing for Media Protection Techniques

cs.CV ·2026-10-16 ·Qiuyu Tang, Aparna Bharati
published
Abstract · 11 pages, 6 figures, accepted by WACV2027

Proactive image protection methods aim to defend visual content against unauthorized manipulation and generative misuse by embedding protective signals into images before release. Evaluating whether these protections remain effective under image editing is therefore critical. However, existing evaluations remain largely limited to one-shot edits, providing only a static view of robustness while overlooking its evolution under realistic multi-step editing workflows. To systematically characterize robustness dynamics under sequential editing across various protection paradigms, we introduce PREEMPT, a holistic evaluation framework. It evaluates protection robustness under both one-shot and sequential editing scenarios, covering image transformations, semantic edits, homogeneous and heterogeneous edit compositions. To enable consistent analysis across protection paradigms with different underlying objectives, PREEMPT normalizes task-specific effectiveness into protection survival trajectories and summarizes long-horizon robustness through trajectory-based diagnostics. Using this framework, we reveal that degradation behaviors remain hidden under conventional one-shot evaluation. Furthermore, perturbation- and watermark-based protection methods exhibit different failure modes, and the first semantic edit largely determines subsequent degradation trajectories. PREEMPT provides a standardized protocol for studying long-horizon robustness and offers new insights into the degradation dynamics of proactive image protection.

0.31

Recognizing Study Lineage Is Not Enough: Over-Crediting Same-Trial Evidence in LLM-Based Medical Verification

cs.CL ·2026-10-16 ·Mukand Krishna, Anoushka Harit, Chenghao Xiao, Zhongtian Sun
code/data
Abstract · 20 pages, 5 figures, 17 tables; 8 pages of main content, with references and appendices thereafter

Publication count is not study count. A model can identify shared trial lineage correctly in a direct task yet over-credit a second publication from the same trial when aggregating evidence. We evaluate this behavior in medical LLM verification using 14 source-audited clinical families and five model conditions. The families span provenance, clinical applicability, result direction, and composition; the provenance families provide the direct lineage tests. Recognition and aggregation are separate stateless calls, and repeated calls assess within-input stability. Across all three original provenance families, GPT-5.5 identifies the independent update in 10/10 recognition calls per family but over-credits the same-trial report in 5/5 separate aggregation calls per family. Selected Qwen3-4B and Ministral-3-8B conditions show the same cross-task contrast. A study-grouping instruction reduces GPT-5.5 over-credit from 4/5 to 0/5 in each later pure-provenance family. Other model conditions show mixed behavior, and rejected outputs limit several comparisons. The results identify a behavioral gap between elicited lineage recognition and evidence aggregation. The code and data artifacts are available at https://anonymous.4open.science/r/study-lineage-review-artifact.

0.30

PONDER: Learning to Generate in the Right Order

cs.CV ·2026-10-16 ·Julius Erbach, Dominik Narnhofer, Bernt Schiele, Jan Eric Lenssen, Konrad Schindler
code/data
Abstract · Project page: https://juliuserbach.github.io/Ponder/

We introduce PONDER, a method that learns in which order an image generator should fill in content and which of its earlier predictions to revise. Masked generative models construct a sample by repeatedly filling in subsets of missing variables. The order of these partial predictions determines the context available in subsequent steps and therefore affects the quality of the final sample. Common strategies are to choose the order randomly or based on local confidence estimates. Conditioned on the partially generated sample and the generator's features, our policy chooses which positions to fill next and/or which previously filled positions to mask for later regeneration, a mechanism we call rethinking. In this way, the generation process is able to adapt to the evolving context and also to revise earlier predictions. We optimize the policy using group-relative policy gradients, with rewards that measure constraint satisfaction, human preference, or distribution quality. For the latter, we introduce FIDgrad, an image-level reward derived from the (first-order) change in Fréchet Inception Distance caused by adding a generated sample to the current feature distribution. With the generators frozen, learned ordering reduces MAR-B ImageNet FID from 4.09 to 2.78 and raises the accuracy on hard visual Sudokus from 3.9\% to 56.6\%, surpassing uncertainty-based greedy ordering (38.7\%). Rethinking further improves these results to 2.70 FID and 96.7\% accuracy, while increasing the rethinking budget of a fixed policy yields continued improvements at inference time.

0.30

StopWatch: Goal Grounding for Object-goal Navigation

cs.CV cs.RO ·2026-10-16 ·Xinhao Liu, Yunsheng Tian, Yuheng Qiu, Pieter Abbeel, Xi Chen +2 more
Abstract · Project website https://stopwatch-nav.github.io

Pretrained VLMs bring rich world knowledge and semantic priors to object-goal navigation, and fine-tuning a VLM into a navigation policy has become a common recipe. Despite this progress, success rates remain low, and it is unclear which capability these fine-tuned policies lack. Our diagnosis identifies goal grounding as the dominant failure. These policies explore well, yet often fail to notice that the goal is already in view or to estimate how far away it is. To address this failure, we ground the goal explicitly. Explicit grounding substantially improves success and reaches state-of-the-art results on six benchmarks across three scene collections. We also correct annotation defects in these benchmarks, so that future methods are evaluated more faithfully. Our analysis on the clean benchmarks shows that, with ground-truth grounding, our policy nearly matches the expert that generated its demonstrations, leaving the headroom in perception rather than policy learning. Our work points to goal grounding as a missing piece of object-goal navigation and offers a simple and effective solution to the task.

0.30

EKT-Router: Expert-Evidence Alignment for Embodied Vision-Language Models

cs.CV ·2026-10-16 ·Weilin Ruan, Yingjie Xu, Yangkai Wei, Bohan Zhou, Shuaibo Li +5 more
Abstract · 15 pages, 5 figures

Embodied vision-language models must answer questions from partial, evolving observations. Their success depends on both the reasoning capability selected and the evidence available to it. Expert routers adapt internal computation, while tool-augmented systems acquire external evidence; choosing these independently can leave a suitable expert without the evidence it needs. We introduce EKT-Router, a bounded framework that jointly selects a capability expert, optional knowledge, and perceptual tools. Outcome-derived signatures represent the overlapping strengths of General, Spatial Reasoning, and Spatiotemporal experts. A joint utility scores complete expert-evidence routes under a call budget. After acquisition, a packet verifier admits valid evidence with positive predicted expert-conditioned gain. The selected expert then answers once, reverting to the unchanged observation if all packets are rejected. Across 6 embodied QA benchmarks, EKT-Router achieves a mean score of 61.4, the highest among the compared systems, and averages 74.7 on 10 general vision-language benchmarks. With all external evidence disabled, question-conditioned expert selection alone improves over the best fixed expert by 1.75 points. Ablations show that learned routing is the dominant component and that knowledge and perception are complementary.

0.30

When Does Evidence-Graph Structure Predict Answer Correctness? Label-Free Abstention for Structured Retrieval

cs.IR cs.LG ·2026-10-16 ·Anjan Goswami
Abstract · 34 pages, 5 figures, 37 tables. Under review at ACM Transactions on Information Systems

Retrieval systems return scored answers, but a high score does not show that an answer should be trusted. The proposed training-free abstention method applies the boundary cut to the selected evidence in an evidence graph. The signal is the weighted cut between selected and unselected candidates and requires no correctness labels or confidence-model training. On MemoryAgentBench (MAB) FactConsolidation multi-hop, with unresolved queries ranked last for every selector, it raises area under the precision-coverage curve from 0.736 for the best score baseline to 0.785 (+0.049, 95% CI [+0.008, +0.092]); under the native evaluation, where score baselines keep their unresolved values, the gap is +0.108. A zero-cut subset contains 71 of 400 queries at 100% observed precision. In this pipeline, the subset corresponds to structural closure; controlled analyses attribute much of the remaining separation to resolution status, with little benefit from continuous edge weights. Across external benchmarks, usefulness varies with graph construction, edge semantics, and relation; typed structure alone does not ensure useful confidence. A supervised gradient-boosted selector achieves 0.814 versus the boundary cut's 0.785 when the retrieval stack matches. These findings provide empirical guidance for using available evidence graphs for lightweight, auditable post-retrieval abstention without additional model inference.

0.30

Rethinking Magnitude Compression in Spectral Losses for DDSP Parameter Estimation

cs.SD ·2026-10-16 ·Byunghoo Park, Jayeon Yi, Minje Kim
code/data
Abstract · 5 pages, 2 figures. Submitted to ICASSP 2027

Spectral losses for differentiable digital signal processing (DDSP) commonly compress magnitudes, for example by using log-magnitude terms. Such compression mimics human perception, but whether it helps the optimization is less clear, and is usually found only by comparing trained models. In this work, we show that whether compression helps depends on the system's informative region: the region of its dynamic range that carries the information pointing toward the target. To indicate where this region lies without having to train, we introduce concordance, which measures how well a loss orders candidate parameter settings by their distance from the target. We then compare uncompressed, compressed, and combined losses by fitting a modal plate reverb and a subtractive synthesizer model to target sounds. A third system, selected for its low-energy informative region, tests the relationship in the opposite direction. In every system, the loss that matches the informative region performs best: the uncompressed one where the region lies at high energies, and the compressed one where it lies low. Surprisingly, the common combination of both log-magnitude and linear-magnitude terms performs best in none of the systems. Code and audio examples are available at https://github.com/superstem000/compression-analysis

0.30

Beyond Multi-Stage Pipelines: A Label-Correlation Ensemble for Multi-Label Retinal Fundus Disease Classification

cs.CV ·2026-10-16 ·Md Asif Uddin, Jannatul Ferdous, Asif Islam, Zobaer Mahmud Zisan, Promit Ghosh +3 more
Abstract

Retinal fundus images frequently carry more than one disease, and public fundus datasets are heterogeneous and severely long-tailed. Clinical triage suggests a staged design, in which a screening gate, a disease-group router and group-specific classifiers operate in sequence, but it is unclear whether this hierarchy earns its complexity. We present a controlled comparison of three architectures on a unified 39-target label space built from RFMiD 1.0, RFMiD 2.0 and ODIR-5K (12,652 images): (i) a three-stage gated cascade of eight models, (ii) a flat model with 39 independent sigmoid outputs, and (iii) a label-correlation model that replaces the flat linear head with 39 learned label queries and a two-layer transformer decoder. All three share one Swin-V2-B backbone, one preprocessing pipeline and one training protocol, so differences are attributable to the architecture. Staging offers no consistent advantage: tracing every missed disease instance to the stage that lost it shows that the two coarse gates discard 34.2% of true instances while the fine-grained classifiers lose only 6.7%, and the gates' hard zeros depress the cascade's macro-AUC to 0.747. Label-correlation modeling gives the largest single gain, raising the detection rate from 0.621 to 0.682 (0.689 with test-time augmentation). Blending the flat and label-correlation models under four-view test-time augmentation is the best configuration on six of eight metrics (exact match 0.694, micro-F1 0.678, macro-F1 0.456, macro-AUC 0.972). On definition-matched class subsets it matches or exceeds four recent studies on macro- and micro-F1, most clearly on low-support pathologies. We also show that ensembling a gated cascade with continuous-output models is harmful, and that per-source negative sampling is needed to remove a dataset shortcut that otherwise yields a spurious AUC of 1.000.

0.30

FlashReFocus: Interactive Image Refocusing in Seconds

cs.CV cs.AIeess.IV ·2026-10-16 ·Ching-Heng Cheng, Chia-Ming Lee, Ming-Ching Chang, Xin Li, Yu-Lun Liu +1 more
benchmarkcode/data
Abstract · 38 pages (8 main + appendix), 47 figures, 16 tables. Project page: https://ming053l.github.io/FlashReFocus/ Code: https://github.com/ming053…

Single-image refocusing changes the focal plane and depth of field of a photograph after capture. It is inherently interactive: users try many focal points and blur strengths on one image, so each edit should be fast and change only the blur. Recent generative methods, however, rerun large diffusion models for every edit. We observe that the two sub-problems differ in kind: recovering the all-in-focus image is ill-posed and needs a generative prior, whereas rendering blur is largely determined by depth and the focal plane. We therefore propose FlashReFocus, which runs diffusion once per photograph and renders every edit with a lightweight network. To address the ill-posed restoration, a one-step diffusion deblurrer with a decoder refiner recovers fine detail efficiently while balancing fidelity and perceptual quality. For precise control, the renderer is conditioned on an explicit defocus map and uses defocus-bias attention, which lets pixels at similar defocus share information without interference from pixels at other depths; in-focus content stays unchanged, and editing the map yields tilt-shift and two focal planes. To further test generalization to unseen cameras, we collect 3CReal, a three-camera benchmark of 89 pairs. FlashReFocus achieves the best perceptual fidelity for zero-shot bokeh rendering on EBB! and 3CReal, refocuses 148.3x faster than GenRefocus, and takes about half a second per further edit, 462x faster.

0.29

Where's the Indirect Evidence? Investigating Grammatical Generalization with Influence Functions

cs.CL cs.LG ·2026-10-16 ·William Dai, Zhenghao Herbert Zhou, Maya Viswanathan, Robert Frank
Abstract · 18 pages

Language models generalize beyond their training corpus to learn linguistic phenomena (Patil et al., 2024; Misra and Mahowald, 2024), showing that there exists indirect evidence driving learning. Prior work has inferred that indirect evidence exists without identifying its character or considered specific hypotheses about what it might be. We explore an alternative bottom-up method for identifying indirect evidence. We use influence functions (Koh and Liang, 2017) to measure how individual training sentences affect model predictions in 15 grammatical phenomena from BLiMP (Warstadt et al., 2020). We filter out a phenomenon's direct evidence using hand-constructed rules and take the 60,000 most influential remaining sentences as indirect evidence. This indirect evidence substantially affects learning: retraining without it lowers held-out accuracy more than removing a matched random group for every phenomenon across five models (by a median of 12 points). We characterize the indirect evidence by its similarity to our direct evidence filters: it (partially) meets the filters' structural conditions at a median of 4.7 times the rate of a matched random group. Finally, clustering phenomena by indirect-evidence overlap recovers three of the five BLiMP categories that include multiple phenomena, suggesting that this method can reveal linguistic relationships.

0.29

Before Retrieval Begins: Safety and Tool-Use Decisions in LLM Agents

cs.CL ·2026-10-16 ·Aditya Nawal, Manit Baser, Mohan Gurusamy
hot direction
Abstract · 19 pages, 5 figures. Preprint version of a manuscript submitted to Transactions of the Association for Computational Linguistics (TACL)

Large language models increasingly act as agents that call external tools, yet most tool-safety research examines only what follows a tool call. The prior decision of whether to call the tool remains largely unexamined and can influence downstream safety. Across the seven open-weight models, the unweighted model-average harm rate after admitted harmful requests is 74.5%. To identify what governs this decision, we apply a factorial intervention to the input request. Within our generated construction, neutralizing the requested-operation component produces the largest admission contrast (19 pp). A linear direction estimated from this contrast influences held-out admission in four of seven models; effects on operation-neutralized requests in some models show that it is not an operation-specific mechanism. We then test whether written refusal is behaviorally and representationally coupled to the same decision. Written refusal accompanies 33.6% of admitted harmful calls. Plain-text refusal training reduces unsafe admission in models that retain benign tool use, showing that response-level safety training can transfer to the call decision. Together, these results establish a partial, model-dependent relation between written refusal and tool admission. They motivate direct evaluation and supervision of tool admission rather than treating text refusal as its proxy.

0.29

QD-Rec: Open-Ended Evolution of Recommendations via Quality-Diversity

cs.IR cs.AIcs.LGcs.NE ·2026-10-16 ·Takashi Ishida, Taro Makino, Issa Sugiura
Abstract

Open-ended search combined with foundation models has produced advances such as faster algorithms, new results on open mathematical problems, and self-improving agents. We extend this paradigm to recommendation for decision-making, focusing on decisions where the set of possible choices is fixed but the reasons for choosing them are not. In high-stakes decisions, e.g., which company to acquire or which customer to target, comparing diverse, well-reasoned options can be more useful than a single ranked list. We formalize open-ended recommendation as building a diverse, high-quality set of recommendations, each an item together with a hypothesis, a natural-language reason why the item fits the needs of a decision maker. We propose QD-Rec, a quality--diversity method that selects items with a contextual bandit and generates hypotheses with a foundation model from earlier ones kept in an archive as stepping stones. A new recommendation enters the archive if it is novel or better than its nearest neighbor. Across M&A, customer targeting, and paper and book recommendation, QD-Rec gives the highest mean quality and builds the largest archive in most settings.

0.29

AMEND: Constraint-Guided Multi-Agent Text Detoxification with Isolated Critics

cs.CL ·2026-10-16 ·Tianyu Wang
Abstract · 10 pages, 2 figures, 4 tables

Text detoxification requires balancing toxicity removal, semantic preservation, and fluency. Existing methods typically optimize an unconstrained scalar objective---enabling safety scores to mask semantic loss---or rely on open-ended iterative debate prone to semantic drift and high computational cost. We propose AMEND, an inference-time framework that decouples detoxification into role-conditioned proposal generation, isolated multidimensional criticism, and constraint-first selection. AMEND evaluates diverse candidates with role-blind critics and enforces strict semantic and fluency feasibility before ranking by utility. To maximize efficiency, AMEND Lite restricts subsequent search to at most one targeted repair round only when residual toxic spans remain. On ParaDetox, one-round selection over diverse roles improves the joint evaluation metric by $+0.0508$ over the best single role. On a frozen confirmation test set, AMEND Lite achieves a joint score of $0.6597 \pm 0.0062$, outperforming unconstrained multi-round loops by $+0.0489$ ($p < 0.05$) while reducing inference requests by 28.4%.

0.29

Benchmarking and Improvement of Artifact Segmentation Methods in H&E-Stained Whole Slide Imaging

eess.IV cs.CV ·2026-10-16 ·Julia Merta, Daniel Mendelsohn, Benedikt Schoser, Joanna Polanska, Michal Marczyk
benchmarkcode/data
Abstract

Technical artifacts introduced during tissue preparation or slide scanning remain a major obstacle to developing robust computational pathology algorithms for hematoxylin and eosin (HE)-stained whole-slide images (WSIs). Despite the growing number of quality-control tools, their real-world utility remains hard to assess because of the limited number of publicly available implementations and the lack of systematic, multi-dataset comparisons. This paper presents a benchmark of artifact segmentation methods for HE-stained WSIs. After a literature review, six publicly available algorithms were selected and compared in terms of their ability to segment artifact-free tissue and selected artifact classes. Two publicly available datasets and a newly introduced dataset released to support further research were used for evaluation. Tissue samples from multiple medical centers and different organs are included, enabling evaluation across heterogeneous data sources. Segmentation performance was assessed using Dice Index, pixel-level recall, and false discovery rate for artifact-free tissue segmentation. For algorithms supporting multiclass artifact segmentation, class-wise Dice Index and inter-class misclassification were evaluated. In addition to segmentation quality, processing time was assessed. Statistical analysis revealed significant differences in segmentation performance between methods. HistoKit, which incorporates a modified version of the GrandQC, was identified as the best-performing algorithm, while HistoQC was the worst. In terms of artifact types, dark spots and out of focus regions were harder to segment than others. Finally, a simple post-processing step was introduced for HistoKit, which further improved segmentation performance. The benchmark establishes a reference for evaluating and developing artifact segmentation methods for HE-stained WSIs.

0.29

What LLMs Do in the Shadows: Reading Context Reliance from the Unembedding Dark Space

cs.CL ·2026-10-16 ·Alexandra Kuleshova, Andrey Volodichev, Daria Kotova, Alexey Zaytsev
hot direction
Abstract

Detecting faithfulness hallucinations in large language models (LLMs) is crucial for their safe deployment in high-stakes domains. Recent approaches address this by contrasting contextual against parametric knowledge utilization, yet measuring context reliance typically demands noticeable overhead, such as materializing attention maps, running additional forward passes, or training auxiliary probes. In this work, we show that context reliance can be estimated by projecting hidden states onto the dark space - directions that compose weakly with the unembedding matrix. Beyond its recognized function as an initial-token attention sink, we show that this subspace actively mediates context processing, and that its share of hidden-state norm - termed the Dark Matter Ratio (DMR) - provides a direct score of context utilization. Pairing DMR with a parametric knowledge score yields DarkLens, a lightweight faithfulness detector operating strictly within a single forward pass. Evaluated across three QA and summarization benchmarks on six open-weight LLMs (4B-13B parameters), DarkLens matches or outperforms a wide range of baselines without requiring extra forward passes or probe training.

0.28

Inferring L1 Backgrounds from L2 English with LLM Token Probabilities

cs.CL ·2026-10-16 ·Ahmet Yavuz Uluslu, Mark Gales, Kate Knill, Gerold Schneider
Abstract

Native language identification (NLI) infers a non-native speaker's first language (L1) from their second language (L2) production. We investigate a token probability approach that scores each text sample as a continuation of a prompt prefix specifying a candidate L1, then predicts the candidate with the highest score. We evaluate our approach using Llama 3.1 8B and 70B on three datasets: TOEFL9, Write and Improve 2024, and BULATS-L1. We compare its performance with previous prompting-based classification approaches. Token-probability scoring substantially improves on direct classification, particularly with the 8B model, while the 70B model achieves state-of-the-art performance across all datasets. Our results indicate that pretrained LLMs' token probabilities contain information useful for distinguishing L1 backgrounds from L2 English production.

0.28

Can This LLM Annotate My Data? Estimating Zero-Shot LLM Performance

cs.CL cs.LG ·2026-10-16 ·Tong Lin, Grant Schoenebeck, Sabina Tomkins
dataset
Abstract

Large language models (LLMs) are increasingly used in annotation tasks with varying success. While several studies have demonstrated the opportunities and shortcomings of LLMs for annotation, a critical question remains unanswered: is it possible to predict how well they will perform on a new dataset? To answer this and related questions, we construct two task families which induce 461 text classification datasets. We find that it is possible to predict LLM performance - to a certain degree - and that contrary to other work which highlights the usefulness of LLM verbalizations of self-confidence, internal log-probabilities are the most predictive signal.

0.28

MGE-PO: Entropy-Aware Credit Assignment for Mathematical Reasoning

cs.CL cs.AI ·2026-10-16 ·Jihao Huang, Guangjia Chai, Zhirui Chen, Lei Wang, Yao Liu +7 more
benchmarkhot direction
Abstract

GRPO assigns the same group-relative advantage to equally rewarded responses within a question, leaving their generation uncertainty unused. We introduce MGE-PO, centered on four-quadrant (4Q) advantage shaping that combines verified correctness with rollout generation entropy. Correctness determines the sign of an additional correction, while entropy controls its magnitude: lower-entropy successes receive larger bonuses and higher-entropy failures receive larger penalties. An online entropy-based question curriculum supports this credit rule using cached rollout measurements. At the final checkpoint after eight training epochs on Qwen3-8B-Base, adding 4Q across three policy optimization settings yields absolute gains of 3.23-3.69% in six-benchmark macro avg@32 and 4.24-11.06% in macro pass@32. On Qwen2.5-7B-Base, the complete configuration raises four-benchmark macro avg@32 from 41.5% to 45.4%.

0.28

Analyzing Repetition and Interference Effects in Memory-Constrained User Simulators

cs.CL cs.AI ·2026-10-16 ·Cara Su-Yi Leong, Nicholas Tomlin, Tal Linzen
benchmark
Abstract

User simulators can help evaluate and train AI systems to interact with humans at scale. To do so, evaluation of user simulators must measure behavioral parity with people. We evaluate whether memory-constrained user simulators are sensitive to two factors known to affect human information retention: repetition of a fact and interference from competing information. We find that some simulators achieve human-like overall accuracy, and replicate the magnitude of repetition and interference effects, but none match people's accuracy on particular questions: while humans misremember facts, models forget them, leading to different error patterns. We release this data as a challenge benchmark for future work on user simulators.

0.28

Best of Both Worlds: Fine-Tuned Small Models Complement LLMs for Cost and Performance Benefits

cs.CL ·2026-10-16 ·Shivam Mathur, Sri Harish Popuri, Priya Krishnan Sundararajan, Kevin Horecka, Michael Mugo
Abstract · Work done in 2024, added for posterity

BERT-based models require thousands of labeled samples to train. Larger models like GPT-X require only 1-5 examples per class. However, performance, cost and time tradeoffs of such models remain under-explored. In this work, we classify survey text data into 16 classes. We focus on experimental approaches and results on performance, cost, and time taken for this problem. We compare these parameters of a trained RoBERTa-based model, and an (off-the-shelf) GPT-3.5 model for text classification. Baseline experiments show that the time taken by GPT-3.5 for generating predictions on 10,000 records is 6 hours, whereas a RoBERTa-based model takes only 12 mins on a T4 GPU. To identify the best prompting approach for GPT-3.5, we explore batch prompting. We experiment with (1) batch-size, and (2) few-shot sampling. We find the best batch size is 20 with human few-shot curation. Experiments show the F1 score of a trained RoBERTa-based model (96%) is much higher than a GPT-3.5 model (83%). In case of GPT-3.5, batch prompting is beneficial in lowering inference costs by ~70%, without hindering task performance. We find the best approach is an ensemble of a RoBERTa-based and GPT-3.5 model (F1: 97.5%) incurring only 4.4% of inference cost compared to a standalone GPT-3.5.

0.28

Recognizing Atari and Assessing Sente: Controlled Evaluation Tasks for Language-Based Go Agents

cs.CL ·2026-10-16 ·Taiyi Chen, Iskar Deng, Amandalynne Paullada
Abstract · 7 pages, 2 figures, 5 tables. The first two authors contributed equally

Domain terminology provides concise descriptions of game states and strategic relationships, but whether game-playing language models can accurately recognize these concepts and relate them to move evaluation remains unclear. We construct a controlled evaluation set centered on atari and its relationship to sente in Go, examining terminology recognition under local board changes and move evaluation under different strategic conditions. We conduct our tests on LoGos, a Go language model, after fine-tuning it with terminology and commentary supervision. The model struggles to recognize changes in the atari relationship and assess its strategic implications. Its terminology predictions are associated with move preferences even when those preferences contradict the reference evaluation. Our controlled tests provide a framework for examining whether language-based Go agents connect domain terminology with precise board relationships and strategic move values.

0.27

In Human Memories. Technology, Creativity and Self-Determination: An Ethical Proposal

cs.CL ·2026-10-16 ·Francesco Vitucci, Giuseppe Silvi, Francesco Abbrescia, Anthony Di Furia, Francesco Scagliola
published
Abstract · 19 pages, 7 figures. English translation of the article originally published in Italian in Ex Chordis, vol. 3 (2024), pp. 15-38. The Italian…

This research investigates the role of artificial intelligence (AI) in modern compositional practices, with a particular focus on two case studies: Convergence (2020/21) by Alexander Schubert, for String Ensemble and AI System, and In Human Memories (2024) by Francesco Vitucci, for four cellos and one electroacoustic player, whose installation was created by the Electronic Music School of the "N. Piccinni" Conservatory of Bari. The initial historical survey traces the evolving use of technology in composition, revealing how the search for new expressive possibilities has driven the development of technological tools through experimentation. The appearance of AI marks a fundamental shift: no longer just a tool, AI now poses questions and challenges to the composer. Engaging with AI becomes an exploration of an unknown entity, whose full capabilities are still largely undiscovered. This new interaction leads composers to confront AI not merely as a passive resource but as a participant in the creative process, outlining an ethical path to reclaim technology as a creative tool. In Human Memories reflects this dynamic, demonstrating how AI can serve as both inspiration and a driving force in shaping contemporary music.

0.27

TAN: A Size- and Form-Adaptive Masked-Convolution Encoder for Tissue Graphs

cs.CV ·2026-10-16 ·Jan Herschel, Anselm von Gladiss
Abstract

Tissue graphs represent histology images as graphs over tissue-conforming superpixels, which keeps homogeneous tissue regions intact instead of cutting them into rectangular patches. To embed superpixels, current encoders crop them to bounding boxes and then pad, resize, or mask them, so background or fill values can leak into the node features. This matters in practice, and especially in medical imaging, where a prediction should depend on the tissue and not on an arbitrary preprocessing choice: on BACH, changing only the fill of the masked region shifts the accuracy of frozen encoders by up to 6 percentage points, larger than the differences among the four best encoders under mean fill, and the preferred fill differs between datasets, so it cannot be fixed once and for all. We present TAN (Tissue-Aware Network), a compact convolutional encoder (0.41M parameters) that processes each superpixel at its native shape and extent and excludes background from every operation, which makes its embeddings exactly invariant to the fill by construction. Under identical masks and an identical graph classifier, TAN's accuracy on BACH is not significantly different from that of the 303M-parameter foundation model UNI (81.1% vs. 80.7-82.6%, depending on the fill), with ~740x fewer parameters, ~1,190x fewer FLOPs, and no pretraining. On BRACS, TAN is close to a ResNet18 trained under the same protocol and trails the histopathology foundation models. An ablation with TAN's architecture but standard processing of fill-padded crops shows that masking keeps accuracy similar (nominally higher on BACH, lower on BRACS, within noise) at less than half the FLOPs.

0.27

GLOBE: Evolving Structured Agentic Memory for E-commerce Product Optimization in Generative Search

cs.CL cs.AI ·2026-10-16 ·Hyunseo Kim, Suil Choe, Sukkyoung So, Daeyeon Yi
Abstract · Accepted to the Second Workshop for Research on Agent Language Models(REALM) at EMNLP 2026

Optimizing product listings for generative search requires language agents not only to rewrite effectively but also to continually self-improve by retaining successful strategies across diverse categories. However, existing approaches often treat structured listings as free-form text, leaving effective strategies implicit in prompts, search trajectories, or rewritten outputs rather than accumulating them as a reusable agentic memory. To address this limitation, we introduce GLOBE, an agentic framework that drives continual self-improvement by storing optimization knowledge in an evolving handbook of explicit, field-scoped behaviors. In each episode, an Editor applies relevant behaviors and receives rank-based feedback, which a Reflector converts into behavior updates. Crucially for dependable execution in real-world settings, a Moderator removes redundant proposals, a Curator commits accepted changes, and consistently harmful behaviors are deactivated. This cycle transforms episodic feedback into a persistent, field-scoped optimization memory. Experiments show that structured behaviors outperform monolithic prompt optimization, while feedback-driven evolution and moderation further improve effectiveness and reduce downside risk. The learned behaviors also remain effective on unseen product categories, suggesting that agentic memory evolution supports interpretable and transferable product optimization.

0.27

HAVANA: A Hierarchical Video-Action Model for Vision-Language Navigation

cs.CV cs.RO ·2026-10-16 ·Dongzhihan Wang, Yi Du, Chen Wang, Xiuxian Li, Liang Xu
code/data
Abstract · 10 pages,6 figures

Vision-Language Navigation (VLN) requires long-horizon instruction-consistent planning and short-horizon local planning in current geometry. Video-action models (VAMs) offer a promising way to learn predictive representations for navigation. However, existing VAMs jointly predict visual futures and local trajectories in a single process, even though they operate at different paces: future prediction remains stable across multiple updates while local trajectories must react to fresh observations. To bridge this gap, we introduce HAVANA, a dual-system VAM that decouples predictive reasoning from reactive local planning. A slow system produces a compact language-conditioned latent plan from observation history, while a fast system combines the latest plan with the current observation to generate local trajectories. Their asynchronous execution enables rapid adaptation without repeated future prediction, and no future video is generated during inference. Experiments on OmniNavBench demonstrate state-of-the-art navigation performance over prior baselines, while real-world deployments further show robust generalization to unseen environments across different robot embodiments. Our code and checkpoints will be available at https://github.com/Anastasiawd/HAVANA.

0.27

MARS: Memory-Aware Routing for Session-Level LLM Assistants

cs.CL ·2026-10-16 ·Minghan Wang, Jinhang Zuo, Zhongxiang Dai, Zhiyong Wang, Fang Kong
Abstract · 32 pages, including references and appendices

Long-term LLM assistants must repeatedly decide what prior context to expose and which model should answer as an interaction evolves. Existing memory systems and model routers usually make these decisions separately, although memory construction changes the evidence, distractors, and difficulty of the resulting input. We introduce MARS, a memory-aware routing framework that uses session state as a shared control interface between memory construction and model selection. A Query and Topic Manager (QTM) estimates compact decision-relevant state that guides a fixed state-conditioned memory policy and an online router, enabling each turn to receive an appropriate memory view and a suitable model under a quality-memory-cost trade-off. We evaluate MARS on RealMem and MultiChallenge with 16 system configurations each, covering long-history memory-driven interaction and bounded sessions with structured context interference. On RealMem, MARS preserves 98.2% of full-context GPT-4o's QA while using only 14.5% of its generation API cost, including both generation and controller calls for MARS, and achieves the strongest measured memory-recall and retrieval scores among compared systems. On MultiChallenge, MARS trails the strongest static system overall but approaches its performance at lower recorded cost on a mechanism-focused subset. Together, these results support a favorable quality-memory-cost trade-off for long-history, memory-dependent interaction.

0.27

Preserving Reasoning Diversity in Reinforcement Learning with Verifiable Rewards via Adaptive Replay

cs.CL cs.LG ·2026-10-16 ·Wen-Zhi Li, Sainyam Galhotra, Claire Cardie
code/data
Abstract

Reinforcement Learning with Verifiable Rewards (RLVR) improves the reasoning capabilities of large language models, but often reduces the diversity of their reasoning behaviors. We study this phenomenon from the perspective of finite-sample learning dynamics and show that limited on-policy rollouts can progressively suppress semantically distinct but equally rewarded reasoning modes. Once a mode becomes sufficiently under-sampled, standard on-policy RLVR rarely recovers it. Motivated by this observation, we propose TAR, a Targeted Adaptive Replay framework that preserves diverse correct reasoning behaviors already present in the base policy. TAR constructs a compact memory bank of diverse base-model solutions and selectively replays them through an adaptive mechanism that activates only when stored behaviors are being suppressed during RLVR. Across six mathematical reasoning benchmarks and two Qwen2.5 model scales, TAR consistently achieves a better balance between reasoning performance and diversity preservation than existing diversity-preserving post-training methods, while substantially reducing the forgetting of correct base-model solutions. Code is available at https://github.com/wenzhilics/tar.

0.27

Prosodic Structure as a Critical Component in Phonemic Language Modelling at BabyLM Scale

cs.CL ·2026-10-16 ·Bill K. S. Ku, Janet B. Pierrehumbert
Abstract · 17 pages, 3 figures. Accepted at the BabyLM Workshop, EMNLP 2026 (archival track)

Human children learn language from continuous, prosodically structured speech long before they can read, yet almost all language models train on orthographic text. Building on previous phonemic BabyLM training (Goriely et al., 2024; arXiv:2410.22906), we ask whether supplying an estimate of the prosody infants exploit improves phonemic language modelling at BabyLM scale. We compare an orthographic baseline to three conditions forming a ladder of phonological richness. All conditions are trained as BERT MLMs at 10M and 100M words. Addressing the ecological validity of existing metrics in the small-data regime, we build a lexical decision task and two semantic organisation tasks. Our new lexical decision task offers improved insights into the distinction between wordlikeness, token familiarity, and lexicality as factors. Triphone phonotactic control and pseudo-log-likelihood scoring shows that these scores reflect phonotactic familiarity rather than wordhood: every model sits near or below chance on an adversarial inverted condition, revealing prior results of this kind as partly artefactual. On our new semantic tasks, a prosodically annotated phoneme stream gives the strongest taxonomic organisation at both the 10M and 100M word scales. Phonemic models with boundary structure are competitive with the orthographic baseline. These findings suggest that prosodic structure is a key component of phonemic language modelling in the small-data regime most analogous to infant learning.

0.27

DinoMUSE: A DINOv2-guided Framework for Unsupervised Lateral Motion Estimation in Ultrasound Elastography

eess.IV cs.CV ·2026-10-16 ·Xiyao Meng, Xingyue Wei, Baihua Luo, Bo Peng, Jianwen Luo
code/data
Abstract

Quasi-static ultrasound elastography (USE) is a non-invasive imaging method for assessing soft tissue mechanical properties. Limited by low resolution, low sampling frequency and lack of phase information in the lateral direction, accurate lateral motion estimation remains a critical bottleneck in USE. Existing methods mainly rely on local image matching and physical constraints (such as tissue incompressibility assumption and effective Poisson's ratio), which lack sufficient contextual information. We present DinoMUSE, a DINO-guided mixture-of-experts (MoE) framework for unsupervised lateral motion estimation in USE. DinoMUSE incorporates high-level visual representations learned by DINOv2 into the motion estimation network to provide global contextual guidance. To facilitate effective interaction between DINOv2 features and local matching features, an MoE-based feature-fusion module adaptively transforms and integrates the two representations. Evaluations against NCC, GLUE, RAFT, and sPICTURE on simulation and in-vivo breast datasets show that DinoMUSE substantially improves lateral estimation accuracy while maintaining competitive axial performance. Specifically, DinoMUSE reduces the mean square error (MSE) of lateral displacement and strain estimation by 60% and 65%, respectively, compared with a SOTA algorithm on the simulation dataset. For in-vivo breast data, DinoMUSE achieves axial strain SNR and CNR of 5.70 and 4.82, and lateral strain SNR and CNR of 4.61 and 2.23, compared to 4.11 and 3.32, and 4.57 and 2.13 for the best competing method, alongside improved lesion visualization. These findings indicate the potential of leveraging transferable visual representations and adaptive feature fusion to improve lateral motion estimation in USE. The code is available at xxxxxxxx.

0.27

MOSS: A Production Framework for Monitoring and Diagnosing Semantic Failures in Search and Recommendation Systems

cs.IR cs.AI ·2026-10-16 ·Alina Nesen, Prathyusha Kanmanth Reddy, David Relyea, Rohit Avinash Mujumdar, Samaneh Karimi +2 more
publishedbenchmark
Abstract · 10 pages. A prior version of this work was presented at the non-archival Unified Search and Recommendation Workshop (USRW), co-located with …

Detecting semantic failures in production recommendation and search systems is hard: behavioral metrics lag by days, manual review does not scale, and most LLM evaluation frameworks are designed for offline, fixed datasets rather than continuous operational monitoring. We present Model Sanity Service (MOSS), a production framework that addresses this operational gap by embedding LLM-based relevance scoring into a recurring monitoring loop with targeted sampling, strategy-aware prompting, structured alerting, and diagnostic follow-up prompting that categorizes failures into actionable root causes. MOSS has been running in production at a large U.S. retailer for over six months across recommendation, search, and offer surfaces, and has surfaced two confirmed serving defects. For one of these, a backfill blending defect, MOSS quantified the impact of the proposed fix offline as a 76.4\% relative reduction in alert rate. On a 498-pair human-annotated recommendation benchmark, frontier LLM judges achieve higher F1 and AUC point estimates than non-LLM baselines (internal production embeddings and SBERT), with the largest gains on strategy-dependent tasks requiring intent-level relational reasoning. The same judge, score scale, and threshold transfer to a 466-pair search benchmark spanning 21 queries with only a surface-specific prompt adaptation, achieving strong F1 and AUC without retuning. These results show that a single LLM monitoring framework can provide useful, actionable quality signals across multiple product discovery surfaces in a large-scale e-commerce setting.

0.27

Can We Use Saliency Maps for the Evaluation of Generative Image Models?

cs.CV ·2026-10-16 ·Eleonora Ristori, Luca Bindini, Marco Lippi, Paolo Frasconi
Abstract

Evaluating generative image models remains challenging, as commonly used metrics depend on fixed pretrained feature extractors optimized for different tasks that do not always align with human realism perception across datasets and model families. To address this limitation, we investigate a human-inspired evaluation framework that avoids representations learned for unrelated tasks. Specifically, we train a generator-specific classifier to distinguish real images from samples produced by a given model and analyze its decisions using saliency maps. Across diverse benchmarks, we observe that these classifiers learn generator-specific artifacts rather than a universal notion of "fakeness", exhibiting strong accuracy and limited cross-generator transfer. This makes saliency analysis particularly informative: the resulting attribution maps highlight visual cues and structural irregularities characteristic of each model's outputs. By combining classification accuracy and saliency information, we obtain a composite score aligned with human judgments, without relying on large pretrained backbones and using only real and generated samples from the evaluated model.

0.27

A low-dimensional dynamical basis for tongue shape control

cs.CL ·2026-10-16 ·Sam Kirkham, Patrycja Strycharczuk
Abstract · Pre-print currently under review

Speech production involves the control of a high-dimensional articulatory system with many degrees-of-freedom, but a widespread hypothesis is that speakers navigate this complexity using a set of lower-dimensional control variables. Common candidates for these control variables include individual articulators (e.g. tongue body) and abstract tract variables (e.g. tongue body constriction location), but identifying the most appropriate coordinate system for articulatory control remains an open question. In this study, we advance a model in which holistic tongue shape modes represent a potential low-dimensional control basis for a simple dynamical law. We first discover empirical tongue shape modes in a cross-speaker coordinate space and then fit the model to ultrasound data on monophthongal vowels from 36 speakers of British English. A three-dimensional linear point attractor fitted to these time-varying modes recovers the dynamics of tongue shape with a high degree of accuracy. This suggests that complex tongue movement can arise from independent modes that linearly combine under simple dynamical laws, thereby relocating some complexity from the control signal to the coordinate system. We show that this framing explains why one-dimensional articulatory measures can exhibit observable nonlinearities, and hypothesize that shape modes are a plausible kinematic expression of underlying muscle synergies.

0.26

Jev-as-a-Judge for Brazilian Portuguese Information Retrieval: Relevance Assessment and Reranking

cs.IR ·2026-10-16 ·Leandro Carísio Fernandes, Jayr Pereira
Abstract

This study investigates Jev, a System One model designed for fast, structured decision-making, in information retrieval. We use Jev-as-a-judge to generate relevance judgments for four Brazilian Portuguese IR collections covering different domains and compare them with the official relevance judgments of each collection. We also evaluate Jev as a reranker. For relevance assessment, exact agreement ranged from 52.9% to 64.9%, while quadratic weighted Cohen's kappa ranged from 0.52 to 0.82. Most disagreements involved adjacent relevance levels, with at least 94.5% of annotations matching exactly or differing by only one relevance level. Despite differences at the individual query--document level, Jev-generated qrels produced similar retrieval system rankings, with Kendall's tau ranging from 0.90 to 0.93 for NormasTCU and Ulysses-RFCorpus across the two evaluation metrics. For reranking, Jev reordered the top 1,000 BM25 candidates for each of the 50 Quati test queries, increasing nDCG@10 from 0.3329 to 0.6331 (+90.2%) and MRR from 0.6852 to 0.9033 (+31.8%). These findings support Jev as a relevance assessor and show improved ranking effectiveness over BM25 in the evaluated reranking setting.

0.26

EduWild: How Teacher-Configured Chatbots Fail in Real-World Use and How to Fix Them

cs.CL ·2026-10-16 ·Ziqi Liu, Ruiwei Xiao, Yuxuan Li, John Stamper
dataset
Abstract

Teachers increasingly configure educational chatbots through natural-language instructions, but how these bots fail in real-world use remains poorly understood. We introduce EduWild, a dataset of 1,726 conversations with 217 teacher-configured chatbots, each paired with the instructions and configuration behind it, including 900 conversations from the intended end users of 35 deployed chatbots. We develop a 29-code taxonomy covering factual, workflow, diagnostic, and pedagogical failures and apply it using a human-calibrated LLM judge. The most frequent failures are lapses in following teacher intent --- bots skip required workflow steps (53.8% of end-user conversations), omit expectation setup (33.4%), and overload learners with information or questions (37.1%). Guided by these findings, we design a prompt-based harness targeting the six most frequent failure modes. In a study with 200 college students across ten bots, the harness reduces the mean number of target failure codes per conversation from 1.406 to 0.489, a 65.2% relative reduction with bots weighted equally. These findings show how real-world failure analysis can guide practical improvements to educational chatbots.

0.26

Discuss More, Comment Less: Boosting Automated Research via Multi-Perspective Collaborative Review

cs.CL ·2026-10-16 ·Junkai Zhou, Jiajia Wu, Zhaoyi Zhang, Shiting Guan, Chuanguang Yang
Abstract

As automated research is increasingly adopted for writing papers, automated review has become a critical link between paper generation and quality control. However, existing methods largely rely on single-model feedback or simple review aggregation, failing to explicitly model expert responsibilities or collective deliberation in actual peer reviews, which limits effective cross-examination and reliable decision-making. Therefore, we formulate paper review as a computational modeling problem of peer review mechanisms and propose RoundtableReview, a deliberative multi-agent framework for enhancing automated research. RoundtableReview structures each paper into eight semantic slots and coordinates six academic and industry reviewers with complementary perspectives, retrieval sources, and specialized responsibilities. Through independent assessment, cross-examination, roundtable deliberation, dynamically weighted voting, and chair synthesis, the framework generates evidence-based and traceable feedback. Experiments on two automated research backbones show that the method improves paper quality in both backbones and outperforms baselines in automatic and human evaluations with fewer comments.

0.26

Spatiotemporal Adaptive Eigenspectrum Remapping in Transcranial Ultrasound Localization Microscopy

eess.IV eess.SPphysics.med-ph ·2026-10-16 ·Stephen A Lee, Isidora Jelenkovic, Maia V. Kokoeva, Jean Provost
unusual
Abstract · 11 pages

Clutter suppression is a critical preprocessing step in ultrasound blood flow imaging, yet conventional singular value decomposition (SVD) filtering depends on data-specific eigenvalue cutoffs that can remove slow flow along with tissue clutter. Here, we introduce adaptive eigenspectrum remapping (AER), a general framework in which clutter is suppressed by continuously remapping, rather than truncating, the eigenvalue spectrum of the spatiotemporal covariance, and we investigate zero-phase component analysis (ZCA) matrix whitening as an instance of AER for transcranial ultrasound localization microscopy (ULM). Rather than discarding eigenspaces, ZCA-based AER whitens the eigenvalue spectrum, reducing the dominance of coherent clutter, including skull reverberation, while retaining weak flow signals. We evaluate AER on both temporal and spatial covariance matrices in mouse brain datasets acquired with long temporal ensembles. Compared with SVD filtering, ZCA improves microbubble contrast and vascular image quality, increases trajectory density across low- and high-concentration regimes, and reveals more low-velocity trajectories in ULM velocity maps. Spatial AER further maintains reconstruction quality across extended ensembles and enables fuller single capillary reporter (SCaRe) reconstructions, including connected artery-to-capillary-to-vein tracks. These results support AER, here instantiated as ZCA whitening, as a data-adaptive alternative to manually thresholded SVD filtering for high-resolution microvascular imaging.

0.26

Disentangling Language Effects on Tool Admission and Post-Retrieval Safety

cs.CL ·2026-10-16 ·Aditya Nawal, Manit Baser, Mohan Gurusamy
hot direction
Abstract · 33 pages, including appendix

LLM agents increasingly use retrieval tools across languages, yet their safety evaluation remains English-centric. It remains unclear whether language affects tool admission, post-retrieval generation, or both. We independently vary request, retrieved-page, and required response languages in a controlled URL-retrieval setting across ten models and seven languages. We identify Safety Loss through Interlingual Processing (SLIP). Non-English harmful requests increase mean tool admission from 46.9% to 67.4%, with a positive mean change in every model. This pattern also persists in an autonomous retrieval setup across all four models tested in this setting. Requiring answers in the page language increases pooled harmfulness on matched tool calls from 50.3% to 57.4% across four closed-weight models, mainly in Chinese and Hindi. Translating pages alone produces no consistent harm increase in eight models with comparable inference settings. Automated audits show that a non-harmful response may be a genuine safety refusal, a deflection citing language difficulties, or an unusable answer. Harmfulness labels alone cannot distinguish these outcomes. Multilingual safety evaluations should therefore measure tool admission and response harmfulness separately and distinguish refusal from capability failure.

0.26

Watermarkability: Measuring and Improving Language-Model Compatibility with Logit-Based Watermarks

cs.CL cs.CR ·2026-10-16 ·Zewei Deng, Mohamed Seif, Liyan Xie
code/data
Abstract · 16 pages, 2 figures, 10 tables. Code available at https://github.com/zeweid1221/LLM_Watermarkability

Language-model watermarks can help identify model-generated text, encode provenance information, and detect post-generation edits. A broad class of logit-based watermarks increases the logits of a designated token set at each decoding step. Because a base model's native probability mass on this set varies across contexts, models, and watermark rules, the same logit bias can yield different adherence rates and increases in perplexity (PPL). We introduce watermarkability to characterize the compatibility between a model and the token sets requested by a watermark. At a target mass $τ$, we quantify this property through a watermarkability cost: the expected minimum KL divergence from the base-model distribution required to assign probability at least $τ$ to the requested set. A lower cost means that the model is more watermarkable for that request distribution. To reduce this cost, we propose a two-stage method. In the first stage, we train a target-set-conditioned low-rank adapter on unions drawn from a 16-way vocabulary partition to increase the requested set's probability mass. In the second stage, a closed-form controller applies the smallest remaining uniform logit bias needed to reach the target mass. We evaluate the method on six watermark schedules using OPT-125M and Qwen3-8B. Across the 12 model-schedule pairs, the pre-adaptation watermarkability cost correlates with the log-PPL increase under a near-hard fixed-bias baseline (pooled Pearson $r=0.929$). For every schedule, at least 99.80% of sampled tokens fall in the target set requested at their decoding step. At this reliability, the adapter and controller reduce average per-schedule prompt-conditional PPL by 46.3% on OPT-125M and 5.2% on Qwen3-8B relative to the fixed-bias baseline.

0.26

Verbalising Model's Computation for Chain-of-Thought Faithfulness Detection

cs.CL ·2026-10-16 ·Yujing Wang, Yulan He, Wenya Wang, Hanqi Yan
Abstract

Current Chain-of-Thought (CoT) verification methods detect reasoning faithfulness from model outputs or static activations, but are limited in explaining how unfaithfulness develops during generation. We introduce VERITAS to verbalise the model's computation: how different reasoning anchors emerge, transfer across layers, and surface in the output. VERITAS adopts the layer-wise Jacobian projection to produce a verbalised account of the model's internal reasoning evolution at each generated CoT token. By training a classifier on the extracted features along the traces, we show that these structured features provide insightful perspectives on how and where reasoning diverges toward unfaithfulness. Experiments on four datasets from FaithCoT-Bench and BonaFide show that VERITAS achieves a significant performance gain over LLM-as-Judge, with a 28.4% gain in balanced accuracy, and an average improvement of 16.0% over activation probing. Further analysis shows VERITAS can be generalised to different datasets and model architectures. Our work shows that verbalising model computation provides a richer and more interpretable lens on CoT faithfulness, moving beyond detection toward a mechanistic understanding of how reasoning fails.

0.25

Same Person, Different Advice: Auditing LLM Financial Advice with Repeated Risk-Tolerance Responses

cs.CL cs.CY ·2026-10-16 ·Soovadeep Bakshi, Apurba Bose, Aishwarya Bose, Sandip Nandi
positionunusual
Abstract · 23 pages, 5 figures, 9 tables

AI financial advisers set a client's stock allocation largely from one stated-risk question, yet people do not answer it the same way twice: in a national US panel, 41% of households chose a different answer two years later, and in a laboratory retest a third of people moved two or more points on a 0-10 scale after six months. Seven large language models from five providers gave one-turn allocation advice for 800 real household profiles in which only the risk answer changed. The recommended stock share moved by 10 to 16 percentage points on average, and by 24 to 35 when the answer changed, against 0.6 to 4.2 points for the same prompt asked twice; in the laboratory retest the rating alone moved advice by 11 to 18 points. Without any note, the models respond to the answer about as strongly as when told that it is very reliable (less so for the two Claude models). If that instructed response is how each model maps attitude to advice, discounting the answer for its measured reliability, as classical test theory prescribes, would cut the swings by a factor of 2.4 to 3.3. The swing is largest for clients under 40. A note that the answer is noisy steadies some models, but shifts their weight to the client's age and pulls cautious clients towards a generic middle allocation: clients unwilling to take any risk are advised 22% stocks instead of 4%. Three extra money questions reduce the swing for some models, with no statistically detectable change relative to how much advice differs across people. In a scripted test, one clarifying question removes nearly all of the swing from an answer the client then corrects (33.5 points to 1.9 on average) while keeping the response to a change the client confirms. We argue that same-person retest audits should be a standard check for AI advisers, and that what to do with an unreliable answer is a question of client autonomy as well as of statistics.

0.25

ProtoGTX: Prototype-Aligned Graph Transformer for Explainable Whole-Slide Image Analysis

eess.IV cs.CVq-bio.QM ·2026-10-16 ·Iori Azuma, Hiroyuki Kusuhara, Tadahaya Mizuno
code/data
Abstract · 14 pages, 3 figures, 2 tables. Code: https://github.com/mizuno-group/ProtoGTX

Deep learning has achieved strong performance in computational pathology, but explaining whole-slide image (WSI) predictions remains challenging. Existing explanation methods often highlight important regions without identifying their morphological patterns, leaving a gap between patch-level computational evidence and slide-level pathological interpretation. We propose ProtoGTX, a graph-based framework that combines spatial contextualization with a flexible morphological prototype vocabulary. ProtoGTX separates the reference prototypes used to define morphological concepts from the predictive model: the reference vocabulary can be generated from training data by unsupervised clustering or supplied externally, and remains fixed during classifier training. Patch features are first contextualized over a spatial graph, after which learnable prototypes representing morphological concepts are optimized through cross-attention. Their slide-level activation profile is aligned with the morphological composition induced by the fixed reference prototype bank. On the CPTAC and BRACS datasets, ProtoGTX outperforms the evaluated MIL, prototype-based, and graph-based baselines, with particularly pronounced improvements on BRACS. Qualitative analyses further show that spatial contextualization transforms fragmented patch-level prototype assignments into more coherent tissue regions. In BRACS, prototype maps also co-localize with pathologist annotations and regions relevant to slide-level prediction. These results demonstrate that ProtoGTX provides complementary semantic and spatial explanations while retaining a prototype vocabulary that can be defined independently of the predictive model. Code is available at https://github.com/mizuno-group/ProtoGTX.

0.25

Perceive, Solve, Reason: Progressive Capability Formation in Time Series-LLMs

cs.CL ·2026-10-16 ·Xin Qiu, Junlong Tong, Wei Zhang, Xiaoyu Shen
benchmark
Abstract

Large language models (LLMs) can now perceive, describe, and reason about images and audio, yet time series understanding remains underestimated: most work uses LLMs as forecasters and classifiers, reasoning studies are few and narrow, and no suite trains and evaluates the full range of capabilities. In this paper, we build the first such suite, with 337K training and 44K evaluation samples across three progressive capabilities, Perceive, Solve, and Reason, and use it to ask how time series should enter an LLM that must master all three, and in what order. We find that a dedicated encoder is essential: with one, a 4B model surpasses GPT-4o on every perception attribute we test, whereas textualizing the series costs 8--10 points in reasoning accuracy. The capabilities then build on one another: learning them in order beats mixing the same data and rivals reasoning-only training with over three times the reasoning data, but only when an encoder is used. Analyses further reveal that capabilities are transferred through distinct network depths rather than simply stacked, and that a stage converged on its own benchmark is not necessarily ready for the next. Distilled into a simple recipe, these findings yield 4B TS-LLMs that, without any target-domain data, outperform four recent 7B TS-LLMs on seven of eight out-of-domain metrics, by up to 11 points. We therefore advocate treating time series as a first-class modality for LLMs, read through dedicated encoders and taught to perceive before they reason.

0.25

How Murky Before a VLM Knows It Cannot See? Accuracy, Confidence and Abstention on a Physical Visibility Axis

cs.CV cs.LG ·2026-10-16 ·Mochamad Rizal Fauzan, Irgi Surya
Abstract · 16 pages, 9 figures, 10 tables

Robustness studies of vision-language models (VLMs) grade degradation on nominal severity scales ("level 1-5") that carry no physical units, and they rarely ask whether a model's confidence follows the loss of visibility. We measure accuracy, internal (token-probability) confidence, verbalized confidence and an explicit abstain option on a visibility axis in metres: fog follows the Koschmieder model, an underwater arm scales the absorption spectrum of pure water, both are applied with a monocular metric depth map, and the visibility range V is swept over eight log-spaced levels from 250 m to 1.95 m. On 1000 COCO scenes and five small open VLMs (2-5 B parameters), accuracy on a boxed-object identification task falls from near its clean value to chance within a narrow band (half-way visibility V_acc of 12.1-15.9 m in fog and 20.5-38.6 m under water) while, for the four models that are confident on clean images, the confidence stays high: at V=1.95 m in fog their accuracy is at chance but their mean internal confidence is 0.68-0.84 and their verbalized confidence 0.61-1.00, a blind-spot width W=log10(V_acc/V_aware) of at least 0.79 decades, i.e. a factor of 6.2 in visibility. The remaining model is under-confident throughout and its internal confidence follows accuracy. An explicit abstain option follows accuracy when it names the reason (W from -0.21 to -0.06 in fog) but, for one model, lags with a generic "I don't know" (W up to 0.18). ImageNet-C fog severities 1-5 correspond to only 55-110 m of physical visibility, outside the failure band, and leave the accuracy of the models within 5 percentage points of its clean value. Blank-image and shuffled-image controls, a prior-matched answer set and a 200-image pilot that preceded the study are reported. Findings are descriptive: one draw of 1000 images, depth of assumed scale, and models of 2-5 B parameters.

0.25

Framing Spatial Hearing within the Free Energy Principle: A Hierarchical Model of Perceptual Learning

cs.SD cs.AI ·2026-10-16 ·Roberto Barumerli, Nils Meyer-Kahlen, Lorenzo Picinali
Abstract

Spatial hearing relies on dynamic internal representations to decode spatial information from ever-changing acoustic conditions. A fundamental example is sound localisation, where listeners infer the direction of an invisible sound. Bayesian inference provides a computational account of such a localisation process; however, modelling efforts are largely limited to static listening scenarios. Instead, the free energy principle (FEP) describes perception, learning, and action as joint minimisation of variational free energy (VFE), providing a principled framework to account for dynamic conditions. Here, we take a first step toward applying the FEP to spatial hearing. We characterise how listeners' beliefs evolve across trials, in which they estimated sound direction and received subsequent visual feedback on each trial. Listeners are described by a hierarchical generative model where visual feedback drives gradient descent on VFE, updating beliefs over three independent quantities: lateral and polar localisation precisions and hemifield disambiguation. The model captures individual learning trajectories and reveals components with different dynamics: lateral precision improves, as does hemifield disambiguation, consistent with increased sensitivity of the auditory pathway in decoding vertical direction. Notably, polar precision also increases, most for listeners with the lowest initial precision, suggesting that the observed vertical improvement is driven by both hemifield disambiguation and polar precision. This work establishes spatial hearing as a tractable domain for FEP and motivates future closed-loop experiments with full action-perception cycles.

0.25

Stabilizing Illumination Conditioning in SSM-Based Low-Light Image Enhancement with Bottleneck Prompt Sidecars

eess.IV cs.CV ·2026-10-16 ·Wei Chen Ng, Qiyi Xie, Jian Ding Tan, Yan Kang Lee, Abdulrab Hakim Qaid Abdullah +1 more
Abstract

Low-light image enhancement (LLIE) requires more than brightness amplification because real low-light images often exhibit spatially varying illumination, amplified noise, color distortion, and local detail loss. Recent SSM- and Mamba-based restorers provide efficient long-range modeling, yet the route for introducing illumination priors into such backbones remains insufficiently studied. This study addresses safe illumination conditioning for SSM-based LLIE through a bottleneck-side PromptAdapter. A lightweight illumination estimator first predicts a single-channel illumination map, which is then converted into prompt tokens and injected through a residual sidecar adapter at the bottleneck, thereby avoiding direct modification of the state-space core. Under a locked three-seed evaluation protocol on LOL-v1, LOLv2-Real, and LOLv2-Synthetic, the PromptAdapter improves over the Base model by 0.94 dB, 1.24 dB, and 0.93 dB in PSNR, respectively, while also improving SSIM, LPIPS, and masked dark-region metrics. The clearest real-data evidence appears on LOLv2-Real. A Learned-Fixed control on LOL-v1 indicates that the present results support the utility of the sidecar conditioning pathway more strongly than a strict claim of illumination-conditioned superiority, and the LOLv2-Synthetic result remains stability-conditioned. These findings suggest that, in SSM-based LLIE, conditioning design can be as important as the conditioning signal itself.

0.25

Variational and Decomposed method for Mitigating Bimodal Bias in Robust Visual Question Answering

cs.CV ·2026-10-16 ·Yihan Zhao, Yuncheng Jiang
Abstract · 10 pages

Visual question answering (VQA) models are known to suffer from the bias issue, which refers to overfitting the regularities of frequent patterns. Previously, researchers only consider the language bias. Recently, the multimodal bias is noticed. However, what multimodal bias is remains unclear. In this paper, we investigate the multimodal bias in counting and non-counting questions. We find that they have distinct underlying mechanism, so their biases come from distinct reasons. We provide valuable interpretations of VQA counting mechanism: \textbf{quantity information stems from the magnitude of the multimodal features in specific dimensions.} Thus, counting bias stems from the vision side and propagates to multimodal features through modality fusion, while the non-counting bias stems from primarily language bias and slightly vision bias. Thus, we propose a method to deal with them differently: Variational and Decomposed method for mitigating Bimodal Bias (VDB). It extracts bimodal bias affected by different proportion of vision and language sides. We introduce variational bias models utilizing the principle of Variational Auto-Encoder to learn mainly language bias with slight vision effect, and a bimodal bias model for bimodal bias combining language and vision sides. Extensive experiments on VQACPv2, VQACPv1, and GQAOOD datasets demonstrate the rationality of our claim and the effectiveness of our method. We also provide explanation for the compromised performance on VQAv2.

0.25

VISTA: A Multimodal Egocentric Benchmark for Goal-Oriented Perception and Assistance to Blind and Low-Vision Users

cs.CV ·2026-10-16 ·Noé Mosca, Mohamed Rayan Barhdadi, Muhammad Hamza, Joshua Yeh, Suryodaya Bikram Shahi +9 more
benchmarkcode/data
Abstract · Code and dataset are available at https://harvard-ai-and-robotics-lab.github.io/VISTA

Blindness and low vision limit many people's ability to complete daily tasks, motivating AI assistants that provide useful guidance. Yet, existing benchmarks do not adequately test whether vision models can provide effective everyday assistance. In particular, i) existing benchmarks primarily rely on visual and textual inputs, with limited support for other modalities that are increasingly available in egocentric wearable devices such as gaze, inertial measurements, spatial trajectories, and others; and ii) existing evaluations emphasize scene understanding and question answering, with limited coverage of goal-conditioned guidance for blind and low-vision users across everyday tasks. To bridge this gap, we introduce VISTA (Visually Impaired Scene and Task Annotations), a multimodal egocentric benchmark for goal-oriented assistance for blind and low-vision users. VISTA contains $\sim$1,000 egocentric recordings captured with Project Aria glasses; each recording provides synchronized sensor streams, six of which serve as model inputs, and three different types of annotation, covering ten task categories (TC), from object localization to hazard detection. We evaluate 14 representative vision-language models zero-shot on VISTA and design a lightweight multimodal model baseline that integrates all six modalities. Experimental results show that, although existing models can often describe egocentric scenes, they remain limited in generating goal-directed assistive guidance. VISTA thus provides a benchmark for evaluating multimodal models that could improve everyday accessibility.

0.25

TriFold: Folding irregular 3D Gaussians into structured latents for 3D Scene Generation

cs.CV ·2026-10-16 ·Yun Wang, Chiyu Wang, Fan Jiang, Ziyu Zhu, Zhaoxu Sun +3 more
code/data
Abstract · 30pages, Project page: https://fantasy-amap.github.io/TriFold

Recent progress in 3D generation has motivated growing interest in compact generative representations of complex 3D scenes. For native 3D Gaussian scenes, however, learning such representations remains challenging due to widely varying primitive counts, nonuniform spatial density, and differences in geometry and appearance even among spatially neighboring primitives. In this work, we introduce TriFold, a native 3D Gaussian VAE that encodes Gaussian scenes into a fixed-layout triplane latent and jointly reconstructs their geometry and appearance. The key insight is to impose structural regularity on the latent space while preserving the spatial adaptivity of the underlying Gaussians. During encoding, we use cell-wise learned queries to aggregate projected Gaussians into multiple slots per cell, reducing the information loss caused by single-feature pooling. During decoding, occupancy prediction identifies active regions, where query-based set prediction and candidate selection reconstruct local Gaussian sets with spatially varying primitive counts. Building on this representation, we train a latent flow-matching model for conditional scene-level 3DGS generation, enabling point-cloud-conditioned synthesis and single-image-conditioned amodal generation. Experiments demonstrate that TriFold achieves high-fidelity reconstruction and generation from a single image that provides only a partial observation of the scene, validating its effectiveness for native 3D Gaussian representation learning. https://fantasy-amap.github.io/TriFold

0.24

Generative Text Modelling and Language-Guided Visual Attention for Deciphering Palimpsests

cs.CV ·2026-10-16 ·Robin Hollifeldt, Tianru Zhang, Jessica Ström, Ekta Vats
published
Abstract · This is a pre-MIT Press publication version, accepted for publication at TACL

Recovering obscured text in palimpsests, i.e. the manuscripts where the original writing (under-text) was erased and overwritten for reuse, is vital for preserving lost information. Due to the nature of the material, such a task can be highly challenging and time-consuming. This paper presents an approach towards automated palimpsest handwritten text modeling for under-text recovery. We investigate generative models, including Generative Adversarial Networks (GANs) and diffusion models, and design a custom denoising diffusion model that learns structured noise patterns to reconstruct the under-text. The recovered text is subsequently processed by a fine-tuned, attention-based handwritten text recognition network. Furthermore, we investigate a language-guided recognition framework that leverages textual class descriptions as informative priors, enabling the text recognition model to distinguish meaningful features from noise. This is achieved by grounding task-specific descriptions, phrased in words or sentences, through visual attention maps generated by pretrained vision-language models like CLIP (Contrastive Language-Image Pre-training). To facilitate future research and benchmarking, we release new datasets for synthetic palimpsests (currently scarce), CLIP-compatible text descriptions, and pre-trained models with code.

0.24

Beyond Surface Cues: Robust LLM-Generated Korean Text Detection

cs.CL ·2026-10-16 ·SungHo Kim, SangKeun Lee
benchmark
Abstract

As large language models (LLMs) generate increasingly fluent text, the need for reliable detection of LLM-generated content has grown substantially. However, existing research remains heavily English-centric, leaving Korean largely underexplored. Moreover, current evaluations of Korean text detection often rely on superficial cues, leading to overly optimistic estimates of detection performance. To address these limitations, we introduce \textsc{KoLing}, a bias-controlled adversarial benchmark comprising 50.9k instances that suppresses superficial shortcuts while preserving semantic content and separately evaluates LLM-generated and LLM-modified text detection. We further propose KoLingProbe, an interpretable detector grounded in theoretical linguistics that models Korean-specific properties across multiple linguistic levels. Experiments show that KoLingProbe substantially outperforms strong baselines on all detection tasks while remaining robust to adversarial attacks. Our analyses further reveal that integrating complementary linguistic features across multiple levels yields both improved performance and interpretability.

0.24

Clinically Informed Acoustic-Aware Attribute Modeling for Respiratory Sound Classification

cs.SD ·2026-10-16 ·Xinhong Li, Li Xiao, Ziyang Chen, Yuhong Yang, Weiping Tu
Abstract · Accecpted by BIBM2026 regular paper

Automated respiratory sound classification (RSC) aims to identify normal and adventitious respiratory sounds, such as crackles and wheezes. Most existing methods formulate RSC as a flat four-class classification problem over norma, crackle, wheeze and both. However, unlike conventional classification tasks with semantically independent and mutually exclusive categories, the commonly used four-class RSC labels have a compositional structure and can be represented by two clinically meaningful sound attributes: crackle presence and wheeze presence. Existing flat classifiers do not explicitly exploit this structure, which may increase the learning burden by treating the four labels as independent targets. Meanwhile, crackles and wheezes exhibit distinct temporal characteristics, with sparse transient crackle events and sustained continuous wheeze patterns, making a single temporal aggregation strategy less suitable for modeling different adventitious sound evidence. To address these limitations, we propose an Acoustic-Aware Asymmetric Framework (AAAF) for RSC. AAAF reformulates four-class recognition as two sound-attribute detection tasks and recovers the final category through logical composition during inference. Built upon a pretrained BEATs encoder, AAAF employs branch-specific multi-layer fusion and asymmetric temporal aggregation to capture localized crackle evidence and sustained wheeze patterns. Auxiliary normal--abnormal supervision and a batch-adaptive dynamic loss are further introduced to improve global discrimination and robustness under imbalanced labels. Experiments on ICBHI and SPRSound demonstrate that AAAF improves overall RSC performance and generalizes across datasets, achieving a state-of-the-art four-class Score of 65.30% on ICBHI.

0.24

Modeling bilingual naming performance through laterally-connected SOMs

cs.CL ·2026-10-16 ·Divya Prabhakaran, Uli Grasemann, Erie Shivers, Marissa Russell-Meill, Swathi Kiran +1 more
Abstract · 9 pages, 5 figures

Computational models have explanatory power to understand cognition in individuals, but there is usually not enough training data to build such models with traditional machine learning. However, alternative architectures that model neural and cognitive processes in more detail can be more easily fit to individuals and can lead to mechanistic insights into human behavior. BiLex is such a computational model of the bilingual lexicon, containing a shared semantic system and phonetic language systems represented through self-organizing maps. In this study, BiLex was extended further into biological realism with excitatory and inhibitory lateral connections. This extension makes it possible to account for response times as well as common errors such as those based on semantic concept familiarity, typicality, and specificity. The Boston Naming Test was simulated in Spanish and English, and model performance approximated naming accuracy and error types in Spanish-English bilinguals with varying primary and secondary language histories. The results indicated that the approach captures semantic-level processes in bilingual lexical access, thus modeling the semantic cognition of individual subjects. In the future, such models may serve as a foundation for modeling impairments and treatment options.

0.24

Do LMs Know What Variables Do? Interpreting Code Representations via Variable Roles

cs.CL cs.LG ·2026-10-16 ·Célian Vasson, Daniil Ognev, Kentaro Inui, Benjamin Heinzerling
benchmark
Abstract · 46 pages, 25 figures

Variable roles describe recurring behaviors such as accumulation, traversal, and retaining the best candidate encountered. We use Sajaniemi's independently defined taxonomy to study which role distinctions can be recovered from large language models, how sparse autoencoders (SAEs) retain those distinctions, and whether associated directions support control of generation. We introduce static annotation rules for all eleven roles and a corpus of 126,926 natural-code files across five languages. Across five model-SAE systems, residual probes outperform individual SAE features and tested lexical baselines, with modest losses after neutral target renaming. Combining SAE features closes much of the single-feature gap, but probes on SAE reconstructions are less sensitive to role-changing edits reviewed by LLMs. High paired ranking accuracy coexists with frequent false positives at fixed thresholds. During generation, interventions often change role annotations at the cost of correctness; changes accompanied by passing tests show no clear aggregate advantage over random controls. The benchmark separates role-related recoverability, SAE fidelity, and correctness-preserving control through matched prediction, source edits, and generation interventions.

0.24

Rethinking the design space of Vanilla Pixel Diffusion Transformers

cs.CV cs.AI ·2026-10-16 ·Soumik Mukhopadhyay, Long Mai, Abhinav Shrivastava, Aniruddha Mahapatra
Abstract · Project webpage: https://soumik-kanad.github.io/mflow-diffusion/. 33 pages, 23 figures, 15 tables. Submission under review

Pixel-space diffusion models have recently re-emerged as a compelling alternative to latent diffusion, avoiding the lossy latent bottleneck and separately trained autoencoder. JiT shows that a vanilla diffusion transformer can generate directly in pixel space by replacing the commonly used velocity prediction ($\boldsymbol{u}$-prediction) with clean image prediction ($\boldsymbol{x_0}$-prediction). In this work, we revisit this choice and introduce MFlow, an adaptive parameterization that learns to interpolate between $\boldsymbol{x_0}$- and $\boldsymbol{u}$-prediction across principal directions of the data, learned end-to-end with diffusion training. We find that the preferred parameterization varies systematically with the data representation and model capacity -- notably, increasing model capacity shifts preference toward velocity prediction. We evaluate MFlow across ImageNet, UCF-101 video, and large-scale text-to-image generation, demonstrating competitive or improved generative performance across diverse settings. In addition, we find that vanilla pixel diffusion transformers, irrespective of parameterization, retain a persistent high-frequency deficit relative to real images. To address this, we introduce ConvRefiner, a lightweight convolutional module trained post-hoc with an adversarial loss while keeping the diffusion backbone frozen. With only a small amount of additional training, ConvRefiner restores the high-frequency statistics of generated images while preserving FID. Project webpage: https://soumik-kanad.github.io/mflow-diffusion/.

0.24

ParetoChat: An LLM-Based Interface for Interactive Multi-Objective Optimization

cs.CL math.OC ·2026-10-16 ·Gokhan Ceyhan
benchmark
Abstract

Multi-objective optimization models are often deployed in settings where decision makers (DMs) must balance some conflicting criteria. In practice, however, stakeholders must translate their qualitative preferences between different criteria into numerical weights, priorities, or reference points required by optimization methods, which can be difficult to specify and interpret. This paper proposes \emph{ParetoChat}, an LLM (large language model)-based expert-agent interface for interactive multi-objective optimization. The agent translates natural-language user input into structured scalarization parameters, invokes an optimization system, presents candidate solutions, and iteratively elicits feedback. We describe a multi-agent architecture that separates input processing, multi-criteria decision-making (MCDM) expertise, solution presentation, and feedback processing. We then present a simulation study using an LLM-based DM agent and benchmark ParetoChat against interactive methods implemented in DESDEO, including reference-point and NIMBUS methods. The computational experiments on multi-objective binary knapsack and set-partitioning instances show that ParetoChat can guide the search efficiently toward solutions that match the DM's preferences. An application to historical courier-order assignment instances illustrates how users can explore objective extremes and intermediate trade-offs through natural-language requests. The results indicate that LLM agents may provide a practical interaction layer between business stakeholders and optimization tools.

0.24

CRISP: Progressive Illumination Curriculum for Recurrent Low-Light Video Enhancement with Semantic Priors

cs.CV ·2026-10-16 ·Ruirui Lin, Xin Yan Lim, Yini Li, Zhiming Liu, Nantheera Anantrasirichai
Abstract

Low-light video enhancement requires the recovery of brightness, color, and fine spatial structures while preserving temporal coherence in the presence of noise and motion. Existing methods often struggle to balance these objectives, particularly in severely degraded regions. This paper presents CRISP, a recurrent framework trained using a progressive illumination curriculum. Its two-frame recurrent architecture propagates information across successive frames to exploit temporal context beyond the immediate input pair. During training, supervision progressively transitions from intermediate targets interpolated between the low-light input and the reference to the full reference, gradually increasing the required brightness and color correction. This curriculum stabilizes recurrent training without introducing additional inference cost and yields the largest performance improvement among the components evaluated in the ablation study. To further support structural reconstruction, the framework optionally incorporates frozen DINOv3 semantic features through adaptive layer normalization (AdaLN)-style modulation, without fine-tuning the foundation model. Experiments on DID and BVI-RLV demonstrate that CRISP achieves the highest PSNR on DID and the highest SSIM on BVI-RLV among the evaluated methods, while maintaining computational efficiency.

0.24

When Evidence Vanishes: Distilling Visual-Evidence Dependency in Medical Vision-Language Models

cs.CV ·2026-10-16 ·Qizhen Lan, Yiling Ma, Ruiqi Chen, Mengchen Fan, Xi Xiao +6 more
theory
Abstract

Compact medical vision-language models (VLMs) can answer clinical questions correctly without reliably using the image, relying instead on question wording or learned priors. Standard knowledge distillation copies a teacher's answer preference on the original image, but it leaves unchecked whether that preference is conditional on the visual evidence required by the question. We study this missing target as medical visual evidence dependency: support for the correct answer should decrease when the matched evidence is absent. We introduce Evidence-Conditioned Dependency Distillation (ECDD), which pairs each training image with a control matched to the evidence level and distills the teacher's drop in answer margin across the pair. A lower bound objective transfers positive dependency by penalizing only shortfalls, while original-image supervision preserves task accuracy. The same formulation applies when the evidence is a queried clinical region, a surgical instrument, or whole-image visual content. Across three medical VQA settings, ECDD keeps original-image accuracy competitive with distillation and alignment baselines while making answer support more sensitive to evidence removal. Under the DME headline objective, for example, the positive-case margin response to target rather than random masking rises from 0.786 with KD to 2.023 with ECDD, while original-image accuracy differs by 0.11 points. These results suggest that medical VLM distillation should transfer both the teacher's answer preference and its dependence on visual evidence.

0.24

Reasoning Reranking with Semantic IDs as the Sole Item Representation

cs.IR cs.CLcs.LG ·2026-10-16 ·Zhenwei Tang, Jintao Liu, Di Bai, Xidong Wu, Xinwu Cheng
code/data
Abstract · 30 pages, 3 figures. Code: https://github.com/lilv98/ReRankID ; data: https://huggingface.co/datasets/lilvjosephtang/ReRankID

Reasoning rerankers write a rationale before ranking a candidate slate, and existing systems read item metadata to do so, even though generative recommenders increasingly represent items as compact Semantic IDs (SIDs). Ranking from SIDs alone has two appeals: SID prompts are far shorter than the matched item text, and although item codes are fixed from content, the backbone learns their token embeddings in part from user-item interaction sequences, so the tokens can carry behavioral signal that item text lacks. That training targets next-item generation, and whether it supports ranking a supplied slate from SIDs alone is untested. We build ReRankID, a controlled testbed in which models reason and rank from a history and slate shown only as SIDs, varying item rendering, history length, and slate size one at a time. On this testbed, such models often copy the input order instead of reranking, and readable metadata does not repair the failure. We compare supervised fine-tuning, rejection-sampled self-training, on-policy distillation, and verifier-based reinforcement learning as post-training methods and find that they elicit the missing behavior. The strongest post-trained model, the 8B model trained with a graded ranking reward, gives the best SID-only ranking and matches strong text-reading LLMs whose prompts are about 20x longer. It ranks above random with no history at all, where a caption-reading reference does not, and it keeps its gains at long histories, where text prompts pass 50k tokens and the caption-reading reference stops improving. In a paired rationale audit, this model argues trade-offs better than that reference but describes items less accurately. The common post-training methods we test thus elicit ranking from SIDs but not faithful item descriptions, which calls for post-training that improves the reasoning itself, not only the ranking it produces.

0.24

SemPurify: Semantics-Aware Data Purification for Black-box Backdoor Defense

cs.CV ·2026-10-16 ·Jiawei Yu, Syed Irfan Ali Meerza, Yi Wu, Amir Sadovnik, Jian Liu
published
Abstract · Accepted by the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2027)

Backdoor attacks have emerged as a critical security threat in machine learning, where adversaries embed triggers within poisoned training samples to manipulate a model's predictions on targeted inputs while leaving its general behavior unaffected. This stealthy nature makes backdoor attacks difficult to detect, as the model functions normally until activated by a specific malicious trigger. Current purification-based defenses use image transformations (e.g., blurring) to disrupt trigger patterns but often degrade image quality. Even with advanced restoration techniques like diffusion models, these approaches can still alter visual or semantic information, reducing utility for downstream tasks. To address this, we propose the first semantics-aware data purification framework for black-box backdoor defense, SemPurify, which explicitly incorporates multi-level semantics (i.e., low-level semantic embeddings and high-level image tags) within a text-to-image (T2I) framework to guide image restoration during purification. Using this semantic information, the T2I model subsequently refines the transformed samples, helping preserve visual details and semantic information during restoration. Extensive experiments spanning multiple backdoor scenarios, i.e., classification models, CLIP-style encoders, and generative models, demonstrate that SemPurify reliably removes backdoors while effectively preserving semantic information for downstream tasks.

0.24

A Model Organism for Studying Unverbalized Evaluation Awareness

cs.CL ·2026-10-16 ·Daniel Donnelly, Evan Chapple, Haritz Puerto, Sahar Abdelnabi
benchmark
Abstract · 63 pages, 4 figures, 30 tables

Large language models can recognize when they are being evaluated and change their behavior accordingly, undermining the reliability of safety benchmarks. Evaluation awareness (EA) is usually measured by having an LLM judge scan reasoning traces for explicit statements of recognition, but this signal is increasingly unreliable. Recent frontier-model system cards report a drop in verbalized rates between releases, which is indistinguishable from awareness that has gone unverbalized. Interpretability methods could detect such unverbalized evaluation awareness (UEA) in individual interactions, but lack instance-level ground truth. Asking the model alters the input, and dataset labels do not reveal whether it recognized a particular prompt as an evaluation. We address this gap with a model organism. Using synthetic document fine-tuning, SFT, and DPO, we train Qwen3.5-27B to emit a marker (a designated token) in its final response when it recognizes an evaluation, while suppressing the marker and verbalized awareness in its reasoning. The marker largely tracks the evaluation concept as it can be strongly modulated by evaluation framing in the system prompt. When not told whether it is being evaluated, the organism recognizes evaluations from test-question cues, mirroring what its base model often verbalizes. Using the marker as a label, we benchmark linear probes, the Jacobian Lens, and Activation Oracles, built on the base model and blind to the marker. None recovers the label reliably. These results suggest that current interpretability methods are not yet reliable UEA detectors, even when recognition relies on simple cues. Our model organism offers a testbed for improving them.

0.23

Reliability-Consistent Appearance Modeling for RGB-Infrared Multi-Object Tracking

cs.CV ·2026-10-16 ·Bingyang Wang, Jiazuo Yu, Lu Zhang, Lihe Zhang, Huchuan Lu +1 more
Abstract · 10 pages, 6 figures. Submitted to IEEE Transactions on Multimedia

Recent RGB-infrared multi-object tracking (MOT) methods have substantially improved cross-modal representation learning and multimodal information integration. However, most existing methods compress RGB and infrared evidence into a unified descriptor. When one modality is temporarily degraded, the tracker can no longer regulate its contribution separately; unreliable evidence may distort association and contaminate appearance memory. We therefore propose RCA-MOT, a reliability-consistent appearance modeling framework that preserves cross-modal complementarity while retaining post-representation evidence controllability. RGB-specific, infrared-specific, and fused identity representations are maintained as jointly accessible appearance states. For each detection, a simple training-free target-scene estimator produces branch-wise appearance reliability used across the two temporal roles of appearance: relative reliability regulates different identity branches, absolute reliability calibrates borderline candidates near the ReID acceptance boundary, and the same branch-wise reliability controls how strongly matched descriptors update historical appearance memory. The individual components are intentionally simple, focusing evaluation on reliability-controllable appearance organization rather than elaborate fusion or learned reliability-estimation modules. RCA-MOT sets new SOTA results on VT-MOT and UniRTL, improving the previous best HOTA by 5.402/6.075 points and IDF1 by 11.549/12.789 points, respectively, while post-hoc analyses further validate reliability as an informative control signal. Code will be released upon acceptance.

0.23

Known Music, Unknown Response: Leakage-Audited Dynamic Arousal Forecasting With Scheduled Future Music

cs.SD ·2026-10-16 ·Bin Wen, Dai-Qiao Zhang, Tien-Ping Tan
Abstract · 17 pages, 3 figures. Code and stored predictions included as ancillary files

Dynamic music emotion recognition usually estimates concurrent or lag-aligned labels from audio that has already been observed. We instead forecast the change in consensus arousal when the upcoming music is scheduled but all later affect labels remain hidden. Evaluation holds out complete songs, encodes each eligible 4-s chunk independently, predicts arousal innovation, and compares the correct future with displaced segments from the same song. On PMEmo and a 1,494-song DEAM evaluation pool excluded from engineering development, a complete past/future MuQ representation raises concordance correlation coefficient (CCC) by 0.389--0.507 over recent affect and remains 0.268--0.382 above a calibrated nonlinear state model. A future-only trajectory contains no past or current audio and still exceeds that state model by 0.150--0.249. The earliest legal future chunk, which ends 5.5--11 s before the target, also remains informative. Correct placement exceeds an overlapping 5-s shift by 0.100--0.209 and the acoustic-nearest eligible same-song segment by 0.078--0.105, with positive song-cluster lower bounds. These results show that both scheduled content and its within-song position carry forecast information under the stated observation rules. The study provides a leakage-controlled reference for known-future music forecasting without treating predictive substitutions as causal interventions.

0.23

Beyond the Benchmark: Conditioning, Domains and Failure Modes of Qwen3-TTS

cs.SD cs.CL ·2026-10-16 ·Séverin Baroudi, Yanis Labrak, Petr Motlicek, Hervé Bredin, Ricard Marxer
benchmarkcode/data
Abstract

Qwen3-TTS reports state-of-the-art zero-shot voice cloning, but its technical report leaves open how the model should be conditioned, how it behaves outside read speech, and how it fails. We characterise its two cloning interfaces, in-context learning (ICL) and X-Vector, against Chatterbox, XTTS-v2 and F5-TTS on read, meeting, emotional and air-traffic-control (ATC) speech, using eighteen objective measures, an LLM judge, a spoofing countermeasure and a listening test. Neither interface dominates: ICL clones timbre more closely from correctly transcribed references, X-Vector clones timing, and ICL degrades sharply when a short reference transcript is wrong. Domain explains four times more variance than system choice. Qwen3-TTS leads on intelligibility and pitch register but trails on prosody, and the objective measures rank systems differently from listeners. An off-the-shelf countermeasure flags its clones no more often than genuine recordings, and on ATC radio it fails for every system. Code and audio are released.

0.23

PreciseEdit: Strong Yet Stable Semantic Transport for One-Step Image Editing

cs.CV ·2026-10-16 ·Jingyi Wang, Xiao Lu
Abstract · 9 pages, 7 figures, and 5 tables

Recent one-step text-to-image (T2I) models enable efficient image synthesis, opening up new possibilities for interactive Web-based image editing. However, existing one-step editing methods often construct a global transport field through simple vector operations between conditional predictions. While such strong transport can effectively drive the image toward the target semantics, it may be released along unreliable directions or over non-edited regions, leading to structural distortions within the editing region and background drift elsewhere. More importantly, stabilizing the editing process through global smoothing or uniformly weakening the transport field often makes large semantic changes overly conservative. In this work, we revisit one-step image editing from the perspective of the tension between semantic strength and transport stability, and identify a key insight: strong semantic transport is not inherently unstable. What matters is whether it is applied along reliable directions and at reliable spatial locations. Motivated by this insight, we introduce PreciseEdit, a model-agnostic, training-free and inversion-free one-step editing framework that strengthens target-semantic alignment while preserving background fidelity and structural consistency. We further propose a dual-noise-scale role decomposition strategy: a higher-noise scale produces the strong target-directed transport used for actual editing, whereas a lower-noise scale jointly serves as a probe for directional agreement and spatial structure. Extensive experiments on PIE-Bench demonstrate that PreciseEdit achieves state-of-the-art performance and effectiveness on various editing tasks.

0.23

FloodVLM-Pix: Language-Guided Pixel-Level Flood Understanding

cs.CV ·2026-10-16 ·Chengyang Zhang, Zhihao Wang, Zhiyu Long, Xueqian Wang, Quanming Yao +2 more
benchmarkcode/data
Abstract

As climate change intensifies, floods are becoming more frequent and destructive, making accurate flood understanding an urgent priority. However, existing vision-language models perform poorly on flood-related tasks, suffering from attentional bias and inaccurate descriptions of flood magnitude. To bridge this gap, we construct the first large-scale remote sensing flood multi-modal benchmark that supports language-level flood understanding and pixel-level flood segmentation. The benchmark contains 131,200 flood image patches and corresponding 443,100 textual descriptions for flood severity and evolution. Based on this benchmark, we propose FloodVLM-Pix, a language-guided pixel-level flood understanding model that includes a flood prior module and a multi-temporal flood pixel decoder to achieve accurate flood understanding. The flood prior module automatically distills sub-region flood spatial priors and inputs priors into FloodVLM-Pix, solving the attentional bias problem in flood language understanding. The flood pixel decoder introduces a novel general-specific feature fusion mechanism to address performance degradation caused by interference (e.g., clouds or mountain shadows) during flood segmentation. Experiments on FloodVL-Pix demonstrate that FloodVLM-Pix improves both flood description quality and pixel-level flood segmentation performance compared with state-of-the-art VLMs and flood segmentation models. Our code and data will be released at https://github.com/Haomiaoyanmeng/FloodPix.

0.23

Prism Attention: Proposal-Refined Index Sharing Mechanism for Efficient LLMs Inference

cs.CL cs.AI ·2026-10-16 ·Zhenxu Tian, Kebin Liu, Zhengwu Yang, Yi Su, Qingqing Dang +6 more
published
Abstract · Published as a main conference paper at NeurIPS 2026

Long-context decoding in large language models (LLMs) is increasingly bottlenecked by the memory and computation required to attend over an ever-growing Key-Value (KV) cache at each autoregressive step. Although token-level sparse attention reduces this burden by restricting computation to a small subset of relevant past tokens, its practical gains are often offset by repeatedly recomputing layer-wise sparse indices as the context grows. A common remedy is to share sparse indices across layers by computing them only at designated anchor layers, but such static reuse ignores the layer-wise evolution of attention patterns, causing the reused indices to progressively deviate from the layer-specific optima and incur substantial attention loss. By revisiting cross-layer sparsity from the perspective of attention dynamics, we find that inter-layer salient token shift is highly localized: although the exact top-$κ$ indices differ across layers, the core high-attention tokens remain stable, with changes concentrated near the expanded boundary of previously selected regions. Motivated by this, we propose Prism Attention, a proposal-refined sparse attention framework. It replaces rigid cross-layer index reuse with a coarse-to-fine strategy, preserving layer-wise adaptability while retaining the efficiency benefits of cross-layer sharing. Extensive experiments show that Prism Attention achieves state-of-the-art performance among mainstream sparse attention methods on both reasoning and long-context benchmarks, while attaining a $3.4\times$ end-to-end decoding speedup over FlashAttention at 128K context length.

0.23

Causal Edit Serialization

cs.CL cs.LG ·2026-10-16 ·Louis Robinson
code/data
Abstract · 37 pages, including appendices and checklist. Code: https://github.com/Louiii/CES ; data and checkpoints: https://huggingface.co/datasets/lo…

Standard autoregressive language models generate text sequentially, but revision requires referring back to tokens in the input context or those already generated. Causal Edit Serialization (CES) is introduced as a framework for expressing such revisions within the same causal token stream. Special pointer tokens are paired with predicted source addresses, allowing insertion, deletion, replacement, copying and movement to be localized explicitly. A lightweight localization head and dual-index rotary position encoding support both address prediction and conditioning on previously emitted pointers, while retaining a standard causal attention mask. A pretraining procedure combines ordinary language modeling with serialized repairs derived from random corruptions, model-generated edits and repeated text. Across model scales and editing baselines, CES models demonstrate complete applied edits with modestly higher clean-text bits per byte than clean autoregressive models. Supervised and reinforcement-learning post-training provide a grammatical-error-correction proof of concept, producing more beneficial than harmful changes under reference-blind model judging. Correct repair programs also improve subsequent text prediction while the original context and its cached states are retained. CES provides a common interface for generation and localized revision, with potential extensions to iterative refinement and additional pointer-addressed transformations.

0.23

Bridge, Distill, Refine: High-Fidelity One-Step Generation for Multi-Bandwidth Speech Extension

eess.AS ·2026-10-16 ·Zhiqi Ai, Xu Shen, Mou Wang, Yongjin Zhou, Shugong Xu
Abstract · 5 pages, 4 tables, 1 fugure

Speech bandwidth extension (BWE) aims to recover missing spectral content from band-limited speech. However, most generative BWE methods are designed for fixed input bandwidths and rely on iterative sampling, limiting their flexibility and efficiency. We propose BDR-BWE, a three-stage generative framework for multi-bandwidth speech extension. First, to enable conditional generation from the observed band-limited spectrum, we design a cutoff-aware conditional flow bridge in the compressed complex STFT domain, focusing generation on the missing high-frequency band while preserving observed spectral information. Second, progressive consistency distillation compresses the multi-step flow transport into a one-step generator. Third, adversarial refinement further improves high-frequency detail and harmonic coherence. BDR-BWE achieves state-of-the-art log-spectral distance (LSD) on VCTK across all evaluated bandwidths with one-step inference, while maintaining competitive perceptual quality and generalizing well to EARS. Audio samples are available at https://aizhiqi-work.github.io/bdr-bwe-demo.

0.23

On the Position Bias of Large Language Models: A Comprehensive Study

cs.CL ·2026-10-16 ·Yining Li, Huayang Li, Lemao Liu, Taro Watanabe
Abstract

Recent studies have shown that position bias remains a critical vulnerability in large language models (LLMs), undermining both the integrity of multiple-choice evaluations and the reliability of retrieval-augmented generation. However, existing understanding of this phenomenon is largely derived from studies with limited model coverage and English-centric benchmarks, leaving the behavior of frontier models and diverse languages underexplored. In this work, we conduct a comprehensive investigation of position bias across model scales ranging from billion-level to trillion-level parameters, covering both multiple-choice and multi-document tasks in five languages. Our experiments reveal that while position bias remains noticeable in small-scale models and Non-English languages, increasing model scale and employing standard Chain-of-Thought prompting significantly mitigate this issue. Moreover, we demonstrate that emerging \emph{thinking models} with extended reasoning capabilities are nearly immune to position bias, suggesting a costly but general approach toward resolving this vulnerability.

0.23

IRAH: Hierarchical Agentic Tree Search for Retrieval Pipeline Optimization

cs.IR cs.AIcs.LG ·2026-10-16 ·Maria Ana Cardei, Aniket Singh, Amir Moghadam, Srividhya Baikadi, Rashmi Shetty
benchmark
Abstract · Under review

Optimizing Information Retrieval (IR) pipelines for Retrieval-Augmented Generation (RAG) requires jointly selecting structural components and numerical hyperparameters under limited evaluation budgets. We introduce the Information Retrieval Agentic Harness (IRAH), which combines Language Agent Tree Search for IR (LATS-IR) with offline meta-learned initialization. LATS-IR performs UCT-guided hierarchical search, first exploring macro-structural pipeline choices and then refining numerical parameters, with natural-language reflections to guide subsequent mutations and provide inspectable search traces. We evaluate IRAH on three in-distribution benchmarks and one out of-distribution benchmark to assess generalization under distribution shift. Under the same meta-learned initialization, LATS-IR outperforms TPE on two of three in-distribution benchmarks, with gains of up to 6.4%, and remains competitive on the out-of-distribution benchmark. Across all four benchmarks, warm-started IRAH achieves the highest average held-out nDCG@10 (0.4816) and improves over its cold-start counterpart. These results establish IRAH as a competitive approach to retrieval pipeline optimization under constrained evaluation budgets.

0.22

Advancing Sentence-Level Dysarthric Audio-Visual Speech Recognition via Test-Time Adaptation

cs.SD ·2026-10-16 ·Shiyao Wang, Hui Wang, Jiaming Zhou, Shiwan Zhao, Yong Qin
Abstract · Accepted for Poster Presentation at NCMMSC 2026

Dysarthric audio-visual speech recognition (DAVSR) enhances performance by integrating complementary visual cues. However, existing research is largely confined to word-level tasks with limited vocabularies, which severely restricts real-world communication for dysarthric speakers. Furthermore, performance remains heavily dependent on supervised fine-tuning with speaker-specific data; significant variations in age, etiology, and severity cause traditional methods to struggle with unseen speakers. To address these domain shifts and the scarcity of continuous multimodal resources, we first introduce a "source speech search-segmentation" pipeline to construct a structured sentence-level corpus from raw CDSD videos. We then propose the use of unsupervised Test-Time Adaptation (TTA) for sentence-level DAVSR using Sequential-level Generalized Entropy Minimization (SGEM). Our evaluations across the state-of-the-art AV-SUPERB and ViSpeR frameworks demonstrate that TTA consistently alleviates Character Error Rate (CER) degradation for unseen pathological speakers. Paired t-tests verify that these improvements are statistically significant (p < 0.05), highlighting TTA as a promising paradigm for advancing speaker-adaptive assistive speech interfaces.

0.22

GeoReason: Bridging Logical Reasoning and Spatial Fidelity in Remote Sensing Segmentation

cs.CV ·2026-10-16 ·Cheng Fan, Han Wu, Tingzhang Luo, Minjing Dong, Yehui Tang +1 more
code/data
Abstract

Multimodal Large Language Models (MLLMs) have significantly advanced instruction-driven perception through the "Embedding-as-Mask" paradigm. However, extending this capability to Remote Sensing (RS) reasoning segmentation remains exceptionally challenging under cluttered geographical contexts and extreme scale variations, often leading to Spatial Fidelity Degradation and Target Misidentification. We identify a critical yet underexplored bottleneck: the sequential alienation of the segmentation token [SEG] from its initial visual grounding within the autoregressive generation paradigm. Produced only at the end of long reasoning chains, [SEG] relies on visual evidence that has been progressively diluted by linguistic abstraction and autoregressive noise. This information decay is particularly catastrophic for RS targets with minuscule pixel footprints, where even marginal fidelity loss leads to localization failure. To address these bottlenecks, we propose GeoReason, a novel framework designed to fortify logical deduction and preserve spatial integrity. First, we introduce an Anticipatory Prior mechanism that injects the [SEG] token directly into the initial user query, shifting it from a terminal output to a primary condition. This paradigm shift transforms the token into a persistent spatial anchor, preventing information decay during long-form linguistic inference. Second, we enhance visual reasoning via a Location-aware Chain-of-Thought (CoT) that enforces coarse-to-fine localization using bounding boxes, coupled with a Latent Patch-to-Token Alignment (PTA) Loss that explicitly aligns the latent segmentation token with physical RS textures. Extensive experiments demonstrate that GeoReason consistently outperforms state-of-the-art methods on the Earthreason and RRSIS-D benchmarks. Code is available at https://github.com/Fanc0126/GeoReason.

0.22

Video Diffusion Avatars: Taming Video Diffusion Models into Controllable 3D Avatars

cs.CV ·2026-10-16 ·Simon Giebenhain, Tobias Kirschstein, Liam Schoneveld, Davide Davoli, Zhe Chen +1 more
benchmark
Abstract · Project webiste: https://simongiebenhain.github.io/video-diffusion-avatars/

Video diffusion models (VDMs) trained on internet-scale data encode a rich prior on appearance, motion, and physics that is difficult to replicate with explicit 3D rendering and costly 3D data capture. We introduce Video Diffusion Avatars (VDAs), a method that creates photo-realistic, multi-view consistent avatars from a single image by fine-tuning a VDM for precise control over camera viewpoint, facial expression, and upper-body pose. Departing from 3D morphable models (3DMMs), VDAs use a learned expression space that lifts the expressiveness ceiling of 3DMMs, e.g., for the tongue and mouth interior. However, learned expression codes exhibit heavy viewpoint bias, as our DISTRACT benchmark reveals. We show that this bias can be removed through disentanglement, enabling their use as driving signals for multi-view consistent avatars. Camera and upper-body pose are controlled jointly via proxy-mesh renderings, giving independent control over expression, pose, and viewpoint. A camera-triplet training strategy makes the approach efficient: our models train for 34 hours on a single A100 using only public multi-view data. VDAs outperform state-of-the-art single-image 3D avatars on novel-view reenactment, reducing face LPIPS by 29% and improving perceptual quality (JOD) by 66% on BecomingLit. A static 3DGS reconstruction probe further confirms that our generations are the most multi-view consistent. Built on a video prior, VDAs produce effects beyond the reach of explicit renderers, including physically plausible hair dynamics, shadows, reflections, and refractions.

0.22

Preservation and loss of algorithmic structure in experimental diffusion of joint-action norms

cs.CL cs.MA ·2026-10-16 ·Dhara Yu, Bill D. Thompson
unusual
Abstract

Everyday norms of social interaction - e.g. norms of sharing space, taking turns, or queuing - are essential to large societies, offering structured procedures for common interactions. However, it remains unclear how the underlying algorithms of these implicit joint plans evolve as they diffuse through populations. Here we show that this process of diffusion can select against complex algorithms in the earliest stages. We conducted a large-scale study of experimental diffusion of novel social norms among human participants, leveraging advances in program synthesis to characterize the procedures people developed in independent populations. Inherited procedures became structurally simplified and lost during interactions between participants who inherited different variants of joint-action norms; however, opportunities for interactive repair offset this pressure, preserving complex algorithms despite misalignment. These results demonstrate that interactive reasoning mediates selective pressures on structural complexity during cultural evolution of procedural norms.

0.22

Behavioral Couplings: From Formation to Behavioral Emergence in Language Models

cs.CL cs.AIcs.LG ·2026-10-16 ·Dongyoon Hahm, Woogyeol Jin, Dennis Wei, Yi Zhou, Nathalie Baracaldo +1 more
Abstract

Fine-tuning large language models (LLMs) can induce behaviors beyond those explicitly targeted. For example, training a model to respond more warmly can make it more sycophantic, and training it to write insecure code can cause broad misalignment on prompts unrelated to coding. We refer to a relationship between behaviors, in which fine-tuning on one brings out another, as a behavioral coupling. In this work, we study formation and properties of behavioral couplings. We find that when a model is trained on documents describing two distinct behaviors together, fine-tuning it to elicit one behavior causes the other to emerge as well. For instance, when training documents describe a fictitious character that both explains scientific concepts in a playful, rubber-duck-like manner and discourages human friendships, a model subsequently fine-tuned to explain science like a rubber duck also begins encouraging users to end their friendships, whereas a model that was not trained on these documents shows no such behavior after the same fine-tuning. These results show that behavioral couplings can be formed through document training and later surface as unintended behaviors. More concerningly, we find that couplings can compose. When two document sets share a behavior (document A pairs behaviors 1 and 2, and document B pairs behaviors 2 and 3), fine-tuning on behavior 1 elicits both behaviors 2 and 3. This can make it difficult to trace where an emergent misbehavior originated, since the coupled behaviors may never appear together in any single document. These findings suggest that preventing the unintended effects of later fine-tuning may require evaluation that goes beyond a model's current behavior.

0.22

LithoWorld: An Endpoint World Model for Lithography Process-Window Recovery

cs.CV ·2026-10-16 ·Botong Zhao, Fang Yu, Xinyuan Chen, Yue Lu
unusual
Abstract

Lithography process drift can move a fabricated pattern outside its process window, producing unqualified scanning electron microscopy (SEM) morphology. We present LithoWorld, an endpoint world model for SEM-guided parameter recovery. Interior representative targets jointly determine qualified reference appearance and correction supervision. A causal predictive branch aligns currentimage features with endpoint representations, while a conditional flow policy generates a two-dimensional correction as an action chunk. The model learns from 173 groups of measured SEMs organized as focus-exposure maps, preserving unmeasured cells as unknown. On 200 test pairs from the 73-group subset, LithoWorld achieves 88.5% qualified-region success and 46.0% interior success, exceeding the strongest comparison method by 2.5 and 6.0 percentage points, respectively. Ablations separate the effects of target selection, endpoint alignment, and output parameterization. Qualitative SEM retrieval illustrates recovery to qualified interiors without requiring exact reference-coordinate matching.

0.22

The Italian lawmaking process: a unified and structured dataset

cs.CL ·2026-10-16 ·Matilde Contestabile, Chiara Ferrara, Giovanni Parrillo, Lorenzo Emer, Andrea Vandin
Abstract

The Italian legislative process is documented in detailed parliamentary records hosted in institutional websites. However, much of this information is available only as unstructured text, limiting its systematic reuse. We applied a controlled, systematic LLM-based extraction procedure, validated against manually annotated records, to convert these texts into ordered and dated events including, e.g., bill introduction, committee examination, plenary examination, and approval. The resulting dataset reconstructs the procedural histories of all 5346 Italian laws enacted between 1985 and 2026, spanning parliamentary terms IX to XIX, for overall 101413 events. Each event records the activity, date, and chamber, with additional procedural details where available. The dataset supports research in the emerging fields of computational legal and legislative studies, potentially combined with process-oriented data science methods to examine procedural paths and legislative timing across parliamentary terms.

0.22

Uncertainty-Aware Budget Allocation for Efficient LLM Tree Search

cs.CL cs.AI ·2026-10-16 ·Jianguo Huang, Zhiyi Lyu, Fuxiang Zhang, Yanchen Deng, Deheng Ye +1 more
Abstract

Tree search enhances multi-step reasoning in large language models (LLMs), yet prevailing methods expand every step with a fixed budget, irrespective of the marginal utility of expansion. We first show, across existing search algorithms, that continuations sampled from a shared intermediate state are largely duplicates and collapse to a few distinct answers. Critically, we find that the semantic dispersion of sibling actions predicts the redundancy of their downstream continuations, permitting a subtree's budget to be fixed before expansion. An information-theoretic analysis motivates this: treating the per-step budget as the Monte Carlo sample size for estimating a step's residual answer entropy, we derive a sufficient budget that shrinks with sibling dispersion. We accordingly propose $\textbf{Uncertainty-Aware Allocation}$ (UAA), a training-free framework that decouples value from uncertainty: values determine which candidates are promising, while uncertainty, estimated from sibling dispersion at no additional model cost, determines how many merit expansion, via uncertainty-aware pruning and uncertainty-aware budgeting. Across mathematical reasoning, logical reasoning, and planning benchmarks, UAA reduces token usage by up to $90\%$ and latency by up to $9\times$ at comparable accuracy, generalizing across four tree search algorithms and both open- and closed-source LLMs.

0.22

Sink-Sparse Hybrid Attention for Efficient Inference in T2I/T2V Diffusion Transformers

cs.CV ·2026-10-16 ·Bingxin Xing, Yu Zhang, Jiatong Ding, Lei Lu, Ping Wang
Abstract · 12 pages, 6 figures

Diffusion Transformers (DiTs) have become a common backbone for high-quality text-to-image and text-to-video generation, but repeated dense self-attention remains a major inference bottleneck as token sequences grow and the same operator is invoked across many denoising steps. Our study suggests that dense attention in pretrained DiTs contains reusable layer/head-specific high-mass structure that can be identified from a small offline calibration set and reused across future generations. This leads to Sink-Sparse Hybrid Attention, a train-free attention replacement that reuses static sink masks and supplements them with a lightweight pooled-KV branch, while leaving model weights and sampling schedules unchanged. On Wan2.1-T2V-1.3B, the method attains 1.54$\times$ end-to-end speedup while maintaining generation quality. On Z-Image-Turbo, it attains 1.57$\times$ end-to-end speedup with competitive GenEval2 performance. These results indicate that reusable importance-aware sparse support can provide an effective train-free acceleration path for both T2V and T2I DiTs.

0.21

Stereo LWIR-RGB Automatic Labeling for Low-Light Camera-Based VRU Detection

cs.CV ·2026-10-16 ·Dimitrios Bouzoulas, Pyry Jäppinen, Pejman Habibiroudkenar, Eerik Alamikkotervo, Risto Ojala
Abstract

VRU detection in low-light conditions is a key component in ensuring VRU safety during the most collision prone time of the day. Currently, advances in low-light detection are limited due to the lack of large scale datasets and automated labeling methods. Most autolabeling methods use RGB detections as their labels source, which cannot be replicated in low-light conditions. To fill this gap in this paper we present an automated pipeline for low-light VRU labeling. We use LWIR-RGB image pairs and source our labels from LWIR as VRUs differentiate from the background due to their body temperature. Unlike our previous work that relied on a complex hardware setup to align the two modalities, we perform this experiment using a LWIR-RGB stereo setup, lowering hardware complexity and cost, making the pipeline easier to reproduce. We evaluate method end-to-end on a dataset collected in the capital region of Finland. Models trained on our generated labels beat both off-the-shelf baselines and models fine tuned on a hand labeled subset on F1, mAP@50 and LAMR metrics when tested on a manually labeled dataset, and the result holds for two different LWIR detectors. This shows that low-light VRU autolabeling can be made both accessible and effective without specialized alignment hardware.

0.21

UniPolar: Joint Image and Polarimetric Restoration under Compound Degradations via Physical Consistency Learning

cs.CV eess.IV ·2026-10-16 ·Yang Jin, Guangyu Guo, Linghao Shen, Binglu Wang
benchmark
Abstract · 13 pages, 11 figures

Polarimetric image restoration aims to recover polarization angle images and reliable polarimetric quantities from degraded observations. Beyond visual fidelity, it also requires preserving the physical relationships underlying reliable polarimetric estimation. However, existing methods typically address either device- or environment-induced degradations in isolation, limiting their applicability to color polarization filter array (CPFA) imaging, where both often coexist. Such compound degradations jointly perturb image intensities and cross angle polarization differences. Consequently, improved image fidelity alone does not necessarily ensure accurate Stokes, DoLP, and AoLP recovery. To address this challenge, we propose UniPolar, a coupled Image Stokes framework for joint image polarimetric restoration from compound-degraded CPFA observations. Specifically, UniPolar employs two complementary streams to reconstruct four polarization-angle images and directly predict Stokes difference components. To coordinate these streams, Degradation Conditioned Cross Domain Interaction combines Dual Domain Frequency Interaction for frequency selective evidence exchange with Degradation Conditioned Modulation for condition adaptive feature correction. Furthermore, at the output level, Image Stokes Physical Consistency Learning aligns image derived and directly predicted Stokes components while regularizing angular fidelity and the physical feasibility of DoLP. On the compound-degradation benchmark, UniPolar achieves average image and DoLP PSNR of 31.522 dB and 24.097 dB, respectively, with an average AoLP error of 16.833 degree. These results underscore the importance of jointly considering visual fidelity and polarimetric reliability, offering a physically grounded perspective on learning based restoration under compound degradations.

0.21

OrTeN: Neural Inverse Graphics for Joint Terrain and Line-scan Camera Trajectory Reconstruction

cs.CV cs.GRcs.LG ·2026-10-16 ·Josef Biberstein, Sertac Karaman
Abstract

Digital terrain maps (DTMs) are used for landing site selection, terrain-relative navigation, and geological analysis in planetary exploration. Existing state-of-practice reconstruction techniques for creating DTMs from satellite imagery use multi-view stereo, requiring complex pipelines for dealing with camera error, changing illumination, and images with limited disparity. This work introduces Orbital Terrain Networks (OrTeN), a DTM construction method that uses neural inverse rendering to jointly optimize a textured height map and orbital camera corrections from line-scan satellite images. OrTeN operates directly on calibrated line-scan images with pixel values expressed as radiance factor (I/F) and corrects both broad trajectory pose inconsistencies and high-frequency pointing jitter jointly with surface reconstruction during training. Surface-projected patch normalized cross-correlation (NCC) enforces consistency between observations of the same terrain, refining an initial rough reconstruction. A semi-empirical reflectance model and shadow rendering are used to fit surface appearance from multi-date imagery with differing solar illumination. OrTeN requires only calibrated line-scan images and initial camera poses and does not use any structural priors including ranging/depth measurements or initial rough DTMs. OrTeN is evaluated on four planetary imagery datasets. Aligned terrain RMSEs are 2.35 m at Jezero Crater, 2.82 m at Gale Crater, 16.18 m at Olympus Mons, and 0.43 m at McGetchin Crater.

0.21

PAC-KD: Pathology-Aware Compositional Knowledge Distillation for Respiratory Sound Classification

cs.SD ·2026-10-16 ·Li Xiao, Weiping Tu, Youqiang Zheng, Dan Chen
Abstract

Respiratory sound classification distinguishes four categories: Normal, Crackle, Wheeze, and Both (crackle and wheeze). Existing knowledge distillation transfers four-class predictions without explicitly exploiting shared respiratory events, potentially limiting the transfer of event-specific knowledge across individual abnormal classes and the Both class. We propose PAC-KD, a pathology-aware compositional knowledge distillation framework that learns crackle and wheeze presence through two dedicated specialists. Their complementary knowledge is transferred to a unified four-class student, enabling it to effectively capture pathology-specific attributes while preserving class-level discrimination. Furthermore, stable conflict calibration adaptively regulates specialist guidance according to its consistency with classification supervision, mitigating potential supervision conflicts. Experiments on ICBHI and SPRSound achieve Scores of 65.28\% and 89.54\%, respectively, outperforming strong knowledge distillation baselines on both datasets.

0.21

Replay Pruning and Few-Bit Integer Inference for Hardware-Efficient Speed-Sign Recognition

cs.CV cs.ARcs.LG ·2026-10-16 ·Se Un Park, Kyungkeon Chung, Huigeol Kim, Junyoung Park
Abstract

Speed-limit sign recognition on an automotive camera system-on-chip (SoC) must fit a detector and a classifier (the reader) into the memory, arithmetic, and energy budgets of the device. In our two-stage recognizer, the detector, a YOLO-family network with ReLU and hard-sigmoid operators, is compressed by channel replay: an iterative pruning cascade identifies the channels to remove, the recorded deletions are replayed on the original task-trained network, the parent, and one continuous knowledge-distillation recovery follows. On a YOLOv5n, replay removes 60.3% of the parameters at the accuracy of the parent and, in validation mAP50-95 over five matched seeds, outperforms iterative pruning and training from scratch, and its per-step variant (per-step replay) is the most accurate pruning method. On a YOLOv8n, replay and per-step replay outperform every other method and the unpruned parent. The reader, a five-layer CNN with 18k parameters, and the post-processing of the YOLOv5n and YOLOv8n heads use integer arithmetic only. With 4-bit weights and 8-bit activations (W4A8), the reader matches FP32 accuracy in 12.1 KB of on-chip memory, and an executable streaming model reproduces the INT8 class predictions of its runtime on all 9,850 test crops. On a Jetson AGX Orin, INT8 is faster than FP16 only in the implicit formulation of TensorRT, and the replay detector runs fastest on the deep-learning accelerator (DLA). Trained on public data only, the combined system of 0.72 M parameters recognizes 95.2% of the signs of a held-out automotive camera at 90% precision and 94 frames per second.

0.21

VISTA: Steering LLMs with Latent User Values Inferred from a Few Survey Answers

cs.CL ·2026-10-16 ·Yinuo Xu, Jadie Adams, Brian Hu, Shivani Kumar, David Jurgens +1 more
Abstract · 8 pages, 4 figures

Personalizing a language model requires knowing what a person values, which stays largely consistent across tasks. Yet most methods either learn task-specific preferences or place the user's profile in the prompt, where models struggle to act on it. We introduce VISTA (Value-Informed Steering and Targeted Acquisition), which infers a latent profile from users' (full or partial) survey answers, and maps the profile through a hypernetwork to a user-specific steering intervention in a frozen LLM to select answer choices or generate free-text responses. Given the same evidence, VISTA matches or beats in-context profiles on synthetic preferences (Food2D), the World Values Survey (WVS), and PrimeX, without adding prompt tokens. To reduce the cognitive burden of completing full surveys, VISTA also chooses which questions to ask and when to stop: on WVS it stops after 6% of the questionnaire while retaining 94% of the full-history gain, where standard adaptive testing retains at most 42%. The dominant steering direction is an interpretable value axis on which VISTA places unseen users accurately. Our results point toward pluralistic alignment that adapts to individuals at deployable cost.

0.21

HydraLLM: Adaptive Width Slicing for Elastic Text and Vision-Language Models

cs.CL cs.LG ·2026-10-16 ·Oscar Kemme, Janek Haberer, Ali Hojjat, Olaf Landsiedel
Abstract · Accepted at the BabyLM Workshop at EMNLP 2026

LLMs and LVLMs are increasingly deployed across heterogeneous hardware, ranging from data-center GPUs to resource-constrained edge devices. Their fixed-size architectures limit the ability to reuse a single model across deployment settings with different compute, latency, and model-size targets. We introduce HydraLLM, an elastic MQA-based architecture. Its number of active attention heads determines the effective widths of the embeddings, attention projections, feed-forward layers, and the output head, forming a nested family of subnetworks within a single parameter set. We introduce a static-to-elastic conversion pipeline that, starting from a pre-trained model checkpoint, includes importance-driven weight reordering and stochastic subnetwork training with knowledge distillation. We extend this principle to multimodal inference through an elastic projection module whose output width follows the active text width. HydraLLM covers a wider range of active parameter configurations than FFN-only elastic architectures. With 75% of heads active, HydraLLM achieves 2.2 percentage points higher TriviaQA accuracy than a state-of-the-art nested baseline configuration which has 17.25% more parameters. With 50% of heads active, our multimodal model achieves a CLIPScore comparable to that of the 7$\times$ larger Gemma 3 4B model. With only 25% of heads active, HydraLLM increases token-generation throughput by 75.3% over the full-width configuration.

0.21

Precision Cotton Disease Management: Leveraging UAV Imagery and Custom YOLO Deep Learning Framework for Enhanced Detection

cs.CV ·2026-10-16 ·1. Anwar Iqbal*1 aiqbal. rime23smme@student. nust. edu. pk 2. Kunwar Faraz Ahmad Khan1 kunwar. faraz@smme. nust. edu. pk 3. Shahbaz Khan1 shahbaz. khan@smme. nust. edu. pk 4. Khawaja Fahad Iqbal1 fahad. iqbal@smme. nust. edu. pk 5. Sara Ali1, 2 sara. baber@smme. nust. edu. pk 6. Saqib Nazir1 saqib. nazir@smme. nust. edu. pk 7. Muhammad Azhar Javed3 muhammadazhar6231@gmail. com 1 Department of Robotics, Artificial Intelligence, SMME, National University of Science +9 more
Abstract

Cotton sustains the textile economies of over 80 countries with an annual yield of 113 million bales, yet 30-60% of yield losses are still attributed to viral, bacterial, and fungal diseases that remain undetected until manual field scouting. To address this critical issue, the integration of Unmanned Aerial Vehicle (UAV) imagery with deep learning offers a scalable early-warning system; however, existing models suffer from low early-stage accuracy, poor transferability, and high computational cost. In this paper, we propose Cotton-Vision, a fully convolutional single-stage detector that synergizes three lightweight architectural innovations: Ghost Bottleneck modules for a 20% parameter reduction, Efficient Channel Attention (ECA) for adaptive feature recalibration, and Cross-Stage Partial (CSP) fusion for enriched gradient flow. The network is trained on a novel dataset collected across seven cotton fields in the Punjab province of Pakistan, comprising over 2,700 high-resolution drone images. Extensive data augmentation (rotation, shear, exposure, and saturation) expands the dataset to 11,704 training samples. On a held-out test set, Cotton-Vision achieves a precision of 97.7%, a recall of 77.7%, an mAP@50 of 85.6%, and an mAP@50-95 of 76.6%. These results represent a 10-20% improvement in precision, recall, and mAP over state-of-the-art baselines, including YOLOv5, YOLOv8, YOLOv11, YOLOv12, RT-DETR, and RF-DETR. Through a comprehensive ablation study, we demonstrate that each proposed architectural enhancement contributes to the model's superior performance, yielding a 12-16% mAP improvement over the baseline architecture. Our findings confirm that this synergy of lightweight and attention-based modules provides a robust, computationally viable solution for early cotton disease detection.

0.21

Momentum: Persistent Dynamics for Video World Models

cs.CV ·2026-10-16 ·Nedko Savov, Stefan Maria Ailuro, Danda Pani Paudel, Yilun Du, Xi Wang +1 more
code/data
Abstract

Video world models must capture how a scene evolves beyond the current camera view to generate consistent observations when its content reappears. Existing approaches rely on observation history, static point clouds, or image caches constructed from past frames. These representations are often expensive and, more importantly, cannot update dynamic scene content outside the field of view: generated vehicles and pedestrians may be forgotten after leaving the field of view or reappear in physically inaccurate states. We represent an evolving scene through synchronized videos from reference-camera rigs, each comprising eight cameras that jointly cover the surrounding scene. We connect these rigs in a graph and propagate video generation between neighboring rigs to coordinate scene dynamics across locations and time. We introduce Momentum, a two-stage pipeline: WorldUpdate jointly generates reference-camera videos as a shared memory of scene dynamics, and WorldInteract grounds camera-controlled video generation in this evolving memory, allowing moving content to remain represented outside the target view and reappear consistently. To support training and evaluation, we develop a CARLA pipeline that captures synchronized videos from reference camera rigs and a moving target camera, together with an evaluation framework measuring consistency within the generated world state and the target observations.Experiments on generated street scenes show that grounding target videos in scene dynamics generated by Momentum improves dynamic scene consistency by 126.2% in unseen scenes over static image conditioning, while also improving visual fidelity. Our models and source code will be made publicly available.

0.21

ASR-Style Transcript Normalization Removes Useful Pragmatic Signal for Dialogue Act Classification

cs.CL ·2026-10-16 ·Lucky Onyekwelu-Udoka, Md Shafiqul Islam
Abstract

Automatic speech recognition (ASR) pipelines often normalize transcripts by removing filled pauses and punctuation and expanding colloquial forms, potentially discarding cues useful for dialogue act classification. We evaluate nine transcript conditions across six classifiers (TF-IDF logistic regression, RoBERTa, BERT, ALBERT, GPT-2, and BiLSTM) and three conversational corpora (Switchboard, ICSI MRDA, and SLUE-HVB), supplemented by multi-seed experiments, normalization controls, and real-ASR evaluations. Across five random seeds, heavy normalization reduces Macro-F1 in all 15 neural model-corpus combinations under the original preprocessing protocols. Removal using the original filler and backchannel lexicon consistently degrades performance on MRDA and SLUE-HVB but produces mixed effects on Switchboard. Punctuation-aware TF-IDF experiments show that punctuation removal reduces Macro-F1 by 0.036 on Switchboard and 0.078 on MRDA, demonstrating that the original linear baseline's apparent insensitivity resulted from tokenization. Restricting filler removal to um, uh, and erm preserves degradation on MRDA but largely eliminates the SLUE-HVB effect, where the original lexicon removes class-defining backchannel expressions. These findings demonstrate that normalization effects depend on preprocessing definitions, feature extraction, and conversational context, supporting task-dependent rather than indiscriminate transcript normalization.

0.21

Beyond the Answer: Student-Conditioned Trace Construction for Reasoning Distillation

cs.CL cs.LG ·2026-10-16 ·Shaoang Li, Yanhang Shi, Yufei Li, Mingfu Liang, Frank Shyu +4 more
benchmark
Abstract

A reasoning trace can accompany a correct answer without providing effective supervision for a particular student. We study how to construct traces that improve learning beyond answer-only supervision by considering both their intermediate content and their accessibility to the target learner. We introduce ReDiST, a student-conditioned trace-construction and distillation framework for evidence-rich reasoning. Starting from the student's own attempts, ReDiST revises answer-correct drafts against the supplied evidence and repairs unsuccessful drafts with additional guidance from the verified answer. A frozen student-matched reader screens and ranks original and edited candidates using answer recovery and verified-answer probability, given the question, evidence, and candidate trace. We distill the selected targets while balancing trace and final-answer supervision with an auxiliary answer loss. Construction and selection occur offline, and inference uses only the distilled student. Experiments on question-answering benchmarks show that ReDiST can improve both source-task and cross-benchmark performance over answer-only, original-trace, and the evaluated trace-selection and rewriting baselines. Across supervision recipes, performance rankings can differ between source and transfer benchmarks, highlighting the importance of evaluating trace teaching utility across distributions.

0.21

FedRAPP: Federated Test-Time Prompt Tuning for Vision-Language Models with Reliability-Aware Prompt Personalization

cs.CV ·2026-10-16 ·Syed Md. Ahnaf Hasan, Md Akil Raihan Iftee, Sajib Mistry, Amin Ahsan Ali, AKM Mahbubur Rahman
benchmark
Abstract · 15 pages, 5 figures and 10 tables

Federated prompt learning enables vision-language models like CLIP to collaboratively adapt lightweight soft prompts across distributed clients without sharing raw data. However, existing federated prompt learning methods are designed primarily for supervised settings with labeled client data and assumed relatively stable training distributions. At test time, source-free federated prompt tuning remains largely unexplored. Clients may receive unlabeled target samples with heterogeneous, often non-IID distributions that differ substantially across clients and from the source training distribution, causing locally adapted prompts to drift in different directions and reducing the effectiveness of federated aggregation. In this work, we explore Source-Free Federated Test-Time Prompt Tuning (SF-FTTPT), where distributed clients collaboratively adapt soft prompts on unlabeled test streams without access to source data. Our study reveals that conventional federated aggregation methods struggle under heterogeneous client distributions, as locally adapted prompts become increasingly specialized and difficult to combine effectively. To address this challenge, we propose FedRAPP, a personalized aggregation framework that combines reliability-aware client assessment, server-side exponential moving-average prompt memory, and similarity-guided peer aggregation. FedRAPP selectively incorporates information from reliable and compatible clients while preserving local specialization, enabling more robust adaptation under distribution shift. Extensive experiments across nine CLIP benchmark datasets and ImageNet variants demonstrate that FedRAPP consistently outperforms standard federated aggregation baselines under heterogeneous client distributions while maintaining the efficiency advantages of prompt-only adaptation.

0.21

Reasoning with Context Management and External Memory

cs.CL ·2026-10-16 ·Man-Ling Sung, William Hartmann, Jiangying Zhou
Abstract · Preprint

Reasoning models maintain information about the status of their reasoning process through in-context memory. Complex problems require larger context windows and memory requirements. In this scenario, context management without caching risks discarding useful information for reasoning. We propose RCMEM, an approach that extends context management with an external memory to maintain additional information without burdening the reasoning model with additional context. RCMEM significantly reduces computation time on problems that require repeated computation of intermediate steps, while achieving better performance on complex problems. We demonstrate our approach on a few mathematics problems using model sizes ranging from 0.4 to 25 million parameters, and GPT-2.

0.21

Location, Location, Generation: Where You Ask Matters for Generative Engine Optimization

cs.IR cs.CLcs.LGcs.SI ·2026-10-16 ·Mahe Chen, Nick Koudas
Abstract · 4 pages

Generative search engines answer local and commercial questions with a written answer and a few cited sources. A growing practice of generative engine optimization (GEO) monitors which sources and brands these answers show, typically from a single place. In this paper, we present a controlled audit of how the user's location changes the sources that generative engines cite. We ask 300 questions of ChatGPT with 19 declared locations, and of ChatGPT and Gemini from five network locations, each twice to measure run-to-run noise (31,200 answers). The engines take location from different channels. ChatGPT follows the location an application declares, down to the city: it writes the place into its own searches, and its search backend localizes results even when the queries name no place. It ignores the network. Gemini ignores the declared location and follows the country of the network, changing most of its sources even for questions that do not depend on place. The businesses the answers recommend change with the sources. Thus, visibility in generative search is specific to the user's location, and audits must control both channels.

0.20

Diversity Without Labels: Task-Agnostic Coreset Selection Across Perception and Action

cs.CV ·2026-10-16 ·Julia Hindel, Benjamin Ramtoula, Kyuhwan Yeon, Nick Heppert, Paul Newman +2 more
Abstract

While data collection for model training is relatively easy, curating high-quality datasets from large collections is inherently challenging. Coreset estimation methods address this problem by identifying the best data subsets to train models on. However, existing approaches often require proxy model training, extensive annotations, or task-specific tailoring. Unsupervised alternatives typically exploit geometric properties in foundation model feature spaces, but are limited to pairwise image-to-image comparisons or comparisons against a fixed subset. We introduce CoreDNA, an unsupervised, task-agnostic coreset estimation approach that pools neuron activations from vision foundation models into per-image and per-dataset distributions. By comparing each image's distribution to the dataset-level distribution, CoreDNA can capture both low-level visual structure and high-level semantic information across layers. We find that high-quality samples concentrate in the tails of the last layer's activation statistics, which we validate by relating them to training-based coreset methods. Building on these findings, we develop a computationally efficient selection mechanism to construct task-agnostic, high-quality coresets for selective annotation and model training. Across image classification, semantic segmentation, depth estimation, and imitation learning for end-to-end driving and mobile manipulation, CoreDNA matches or outperforms the state of the art, demonstrating unprecedented generalization across perception and action tasks. These results highlight the effectiveness of compact activation-based representations for data selection.

0.20

Steering beyond Language: Improving Multilingual Training-Free Text Embeddings via Paired-Residual Subspace Removal

cs.CL ·2026-10-16 ·Ailiang Lin, Zhuoyun Li, Yusong Wang, Keyu Mao, Kotaro Funakoshi +1 more
Abstract

While deriving text embeddings directly from frozen LLMs has emerged as a promising direction, existing methods focus almost exclusively on English, leaving multilingual settings largely unexplored. In this work, we present the first systematic study of multilingual training-free text embeddings in LLMs. Our analysis reveals a language-conditioned embedding geometry in which semantically equivalent sentences remain misaligned across languages. We hypothesize that language-specific variation can be disentangled from semantic content to improve embedding quality. To this end, we propose Paired-Residual Subspace Removal (PRSR), a training-free inference-time intervention that steers LLM representations toward a more language-invariant semantic space. PRSR first constructs paired residuals from multilingual parallel sentences to capture language-specific deviations. It then estimates layer-wise low-rank subspaces from these residuals and suppresses the corresponding components in the last-token hidden states. We conduct more than 200 monolingual evaluations across 18 languages using MTEB and MMTEB, along with cross-lingual bitext mining between English and over 200 other languages. Empirical results demonstrate that PRSR consistently improves the performance of existing training-free baselines in both monolingual and cross-lingual settings across different LLMs.

0.20

Semantic-Critical Regions for Video-to-Audio Generation

cs.CV cs.MMcs.SDeess.AS ·2026-10-16 ·Yun Fang, Wei Guo, Zhuonan Liang, Heng Wang, Weidong Cai
published
Abstract · 11 pages, 7 figures, 7 tables, accepted by ACM Multimedia Asia 2026

In real-world videos, sound events are often associated with localized and fine-grained visual cues, such as object motion, object-part dynamics, contact events, and interactions between objects. Existing video-to-audio (V2A) methods employ implicit conditioning based on global visual representations or explicit control derived from temporal and spatial cues. However, subtle sound-producing and sound-indicative interactions remain insufficiently captured, leaving room for more precise visual conditioning of sound-event semantics and temporal dynamics. Motivated by this observation, we refer to regions containing such sound-relevant visual evidence as Semantic-Critical Regions (SCRs) and introduce SCR-V2A, a V2A generation framework that explicitly localizes and encodes SCRs as fine-grained visual guidance for audio generation. Specifically, we first identify SCRs through sound-related visual localization. We then transform the localized regions into a control representation through adapter-based dual visual encoding, which jointly captures global visual context and fine-grained sound-relevant evidence. The representation is used to condition a downstream audio generation model. Experiments on both in-domain and out-of-domain benchmarks demonstrate that SCR-V2A achieves strong performance in distribution matching and audio quality while remaining competitive in cross-modal alignment and requiring few trainable parameters. These results highlight the value of prioritizing fine-grained, sound-relevant visual regions for V2A generation.

0.20

SLA-LENS: From Classroom Transcripts to CSLA-Informed and Evidence-Linked L2 Interaction Analytics

cs.CL ·2026-10-16 ·Nemika Tyagi, Olga Kellert, Phanos Kareen, Nelvin Licona-Guevara
Abstract · 16 pages, 3 figures, 14 tables

What happens after a learner makes an error, and how does the instructor's response shape the next turn? Investigating these exchanges across classrooms requires context-sensitive annotation that is difficult to scale manually. We introduce SLA-LENS, an end-to-end LLM-assisted framework that connects classroom transcripts to structured pedagogical annotations and evidence-linked descriptive analytics. Grounded in interactionist Second Language Acquisition (SLA) theory, its annotation schema captures second language learner errors, pedagogical functions, and uptake or repair while accounting for speaker roles, conversational context, and hierarchical labels. We apply the framework to 40 authentic online L2 Spanish classroom transcripts comprising over 21,000 sentence-level instances. We evaluate proprietary and open-weight LLMs, including task-specific QLoRA adaptation, using expert-reviewed gold annotations, held-out silver data, and metrics sensitive to different types of annotation error. A deterministic reporting layer generates reproducible visualizations of participation, feedback, learner errors, uptake, and temporal interaction patterns, linked to supporting transcript evidence. By making these patterns accessible and inspectable, SLA-LENS supports systematic classroom discourse research, pedagogical reflection, and curriculum design.

0.20

STRIDE: Evaluating Spatiotemporal Reasoning in Driving Edge Cases

cs.CV cs.AI ·2026-10-16 ·Lieqi Liu, Rui gao, Jia-Chen Gu, Wenbo Hu, Zhaobin Mo +4 more
benchmarkcode/data
Abstract · Project page: https://pluslabnlp.github.io/STRIDE/

In autonomous driving, perceiving surrounding objects is not enough: failing to understand how they move and interact over time can lead to dangerous decisions. Evaluating vision-language models (VLMs) therefore requires testing whether they connect past spatial observations to future consequences. Yet existing benchmarks often assess edge cases through basic perception questions or test temporal understanding in simple scenarios, leaving complex spatiotemporal reasoning underexplored. We introduce STRIDE (SpatioTemporal Reasoning In Driving Edge Cases), an object-centric benchmark for long-tail, safety-critical driving scenarios. STRIDE covers six task families: spatial perception, spatial understanding, temporal memory, temporal extrapolation, trajectory prediction, and scene-context awareness. It pairs multiple-choice questions probing individual spatiotemporal judgments with related free-form visual question answering (VQA) requiring detailed scene explanations and justifications of driving decisions. We prioritize scenes with unusual object behaviors and complex traffic interactions, generate reference answers from structured dataset annotations through automated procedures, and use limited language-model assistance and human verification to ensure annotation quality. Across 12 proprietary, open-weight, and driving-specialized models, even the best performer, GPT-6-Astra, achieves only 43.4% average multiple-choice accuracy, versus an approximately 20% random-choice baseline. Further analysis reveals that strong performance in describing driving scenes does not necessarily translate to accurate safety-critical judgments, and that models struggle more with reconstructing object motion than recalling past positions. STRIDE thus exposes spatiotemporal driving limitations that aggregate VQA scores can conceal.

0.20

Language Models are Lazy Poets: Creative Mode Collapse at Scale

cs.CL ·2026-10-16 ·Jacob A. Matthews, Sil Hamilton
Abstract

While aligned large language models (LLMs) can successfully complete many creative tasks, their outputs tend to be highly self-similar. Though solutions to this problem of "mode collapse" have been proposed, comprehensive large-scale surveys of the phenomenon in specific textual genres are still lacking. We address this gap through an extensive treatment of LLM mode collapse in poetry. Across 15 poetic forms, 248 subject matter tags, and 16 open-weight models, we generate over 1.1 million poems. To our knowledge, this constitutes the largest corpus of LLM-generated poetry to date. At million-poem scale, we find that LLM poetic vocabularies and rhyming abilities are substantial, but severely underutilized. In other words, LLMs are "lazy" poets: despite possessing large poetic inventories, LLMs over-rely on a small number of words, opening phrases, and rhymes. Further, we find that lexical homogenization differs across models and poetic forms, noting observed rhyming vocabulary is correlated with model size.

0.19

Sentence-Level Topic Modeling with Lexical and Contextual Evidence

cs.CL cs.LG ·2026-10-16 ·Federica Granese, Serena Villata, Charles Bouveyron
Abstract

Topic modeling faces a fundamental granularity dilemma: probabilistic topic models assign topics at the word level, whereas transformer-based methods operate at the document level. The latter overlooks that real-world documents typically cover multiple topics, whereas the former relies on individual words, which rarely provide sufficient context to identify them. We introduce SentenceETM, a neural topic model that assigns a topic to each sentence by jointly using lexical and contextual evidence. Each topic is described both by a distribution over the vocabulary, defined through word embeddings, and by a Gaussian over sentence embeddings concentrated on a topic-specific low-dimensional subspace. The topic of a sentence generates both its words and its embedding and is marginalized exactly, so that its posterior combines the two sources of evidence. Document proportions are inferred with amortized variational inference, while the Gaussian parameters are re-estimated in closed form. Experiments show that SentenceETM consistently improves sentence-level topic assignment over state-of-the-art approaches, and a case study of a presidential debate shows how sentence-level topics reveal how candidates handle the same subjects.

0.19

Prune the Leaky Neurons: Mitigating Private Data Leakage in Fine-Tuned LLMs via Joint-Embedding-Guided Pruning

cs.CL cs.AI ·2026-10-16 ·Mohamed Shaaban, Mohamed Elmahallawy
Abstract

Fine-tuning large language models (LLMs) on proprietary records enables domain adaptation but introduces two challenges: (i) privacy leakage of memorized private data, such as personally identifiable information (PII), and (ii) balancing private data removal with utility preservation. Existing defenses, including data sanitization and differential privacy, often degrade performance, while privacy-neuron editing lacks a PII-free reference to distinguish identity-specific neurons from those supporting useful text. We propose JEPA-Prune, a post-hoc neuron pruning framework guided by a model trained exclusively on PII-free views. Alongside the fine-tuned private model, JEPA-Prune, trains a public model on surrogate and masked views of the same records using a joint-embedding predictive architecture (JEPA), capturing domain knowledge without identifiers. A cross-model representation probe quantifies deviations between private representations of leaked records and this PII-free reference. Integrated gradients and a scale-invariant specificity score then identify PII-responsive neurons, which are pruned without retraining. Experiments show that pruning just 1.6% of neurons reduces PII inference accuracy by up to 7.4x and PII extraction recall by up to 13.9x, while keeping unseen-text perplexity within 16% of that of the unpruned model for most LLMs. In contrast, privacy-neuron editing increases perplexity by at least 26x.

0.19

Acceptance Is Not Speed: Length Inflation in Training-Free Lossy Speculative Decoding

cs.CL ·2026-10-16 ·Chiatzen Wang, Bill Xu, Haochen Gu, Hongyang Zhang
Abstract · Submitted for NAACL 2027, under review

Speculative decoding typically characterizes speedup by draft tokens accepted per verification round. Lossless verification preserves the target model's output distribution and thus expected completion length, making this metric a meaningful efficiency proxy. Relaxed (lossy) verification accepts tokens that lossless verification would reject, potentially altering that distribution. We identify a length inflation phenomenon: relaxed verification lengthens completions, primarily by extending reasoning traces, largely offsetting apparent gains in accepted tokens per round. We evaluate existing training-free relaxed verification rules on twelve target-drafter pairs against lossless verification on identical prompts, measuring completion length, verifier rounds, wall-clock time, and task accuracy. At each rule's most relaxed evaluated setting, relaxed verification on average accepts $1.34\times$ as many tokens per round but produces $1.54\times$ longer completions. For rules with higher acceptance lengths, tokens accepted only under relaxed verification tend to have low target-model probability, contributing to length inflation. For rules that reduce verifier rounds, accuracy typically drops substantially. Further analysis traces some failures to continued reasoning that exhausts the budget needed to deliver a final answer. Overall, accepted tokens per round alone is an unreliable measure of decoding speed under relaxed verification. Completion length and verifier rounds must also be reported to assess actual efficiency.

0.18

GliomaINK: A Tumor and Anatomy Conditioned Flow Matching Framework for 3D MRI Sequence Synthesis in Glioma Patients

eess.IV cs.CVq-bio.TO ·2026-10-16 ·Moinak Bhattacharya, Annie Singh, Mohammad Chaudhry, Prateek Prasanna, Gagandeep Singh
Abstract

Background: Incomplete multiparametric MR examinations limit glioma characterization and downstream AI tools. Generative models can synthesize missing MRI sequences, but anatomical structure and clinically relevant tumor features are challenging to preserve. We propose GliomaINK, an anatomy-guided conditional flow matching framework for cross-sequence 3D brain MRI synthesis. Methods: A 3D SwinUNETR predicted the conditional velocity field while an auxiliary topology-preserving loss encouraged structural consistency. The model was trained on 1,083 patients from the BraTS-AG dataset and independently tested on an in-house clinical cohort. Downstream tasks include image quality, tumor segmentation, IDH1 mutation-status prediction, and overall survival prediction. Two radiologists independently evaluated the anatomical fidelity, pathological representation, and overall image quality of the synthesized images. Results: GliomaINK exhibited the highest PSNR across all six MRI translation directions in both evaluation cohorts. In the BraTS-AG test set, the PSNR reached $25.64\pm0.55$ dB for the T2W-to-T1C synthesis, compared with $13.78\pm0.32$ dB for Pix2Pix and $18.90\pm0.53$ dB for SPADE. For downstream tumor segmentation, the method showed strong performance across tumor subregions with Dice scores reaching $0.928\pm0.014$ for edema, and $0.885\pm0.040$ for enhancing tumor. For the IDH1 mutation-status prediction, GliomaINK achieved the highest AUROC score in four of the six sequence configurations, reaching $0.827\pm0.088$. Conclusion: Anatomy-guided conditional flow matching enables high-fidelity cross-sequence glioma MRI synthesis demonstrating potential for preservation of clinically relevant information in synthesized MRI and improving downstream tasks.

0.18

VTrajEmbed: Aligning Text with Video Object Trajectories via Spatiotemporal Embeddings

cs.CV ·2026-10-16 ·Ruizhi Chen, Zining Fan, Yudi Guo, Jiang Liu, Zhiyu Lin +3 more
benchmark
Abstract · 28 pages, 1 figure

Fine-grained alignment between text and video object trajectories is a core challenge in multimodal understanding. Existing embedding models excel at global retrieval, yet their single global representation entangles multiple objects and their spatiotemporal dynamics, limiting direct object-level grounding. To address this issue, we propose VTrajEmbed, an embedding model that decomposes a video into object trajectories and aligns each trajectory embedding with text for object-level grounding through retrieval. Our main contributions: (1)Slow-Fast Trajectory Encoding, a dual-branch architecture that establishes trajectory representations by densely sampling low-resolution frames for high-frequency motion cues and sparsely selecting high-resolution frames for scene context, aggregated by a shared LLM backbone; (2)Mask-Guided Contrastive Learning, a training objective built upon these representations that couples an intra-group loss and a batch-level bidirectional loss under a relation mask, excluding unverified relations and preventing confirmed matches from becoming false negatives; (3)Text2Traj-Bench, a trajectory-level text-video benchmark with model-assisted construction and human-verified query--trajectory relations. Experimental results show that VTrajEmbed-5B improves text-to-trajectory and trajectory-to-text R@1 over Qwen3-VL-Embedding-8B by 15.48 and 18.05 percentage points, respectively, and achieves 49.48% average R@1 across five global retrieval benchmarks, compared with 42.86% for Qwen3-VL-Embedding-8B.

0.18

Query-adaptive Recurrent Tokenization

cs.CV ·2026-10-16 ·Bavesh Balaji, Pranav Gopinath, Pouya Sadeghi, Sirisha Rambhatla
Abstract

Visual information worth extracting from an image and the token budget it warrants depend on the text query. Yet modern vision and vision-language models largely encode images independently of the query and allocate a fixed visual token budget to every input. We introduce the Query-Adaptive Recurrent Tokenizer (QART), a backbone-agnostic module that makes visual representations both query-steerable and budget-adaptive. QART steers the visual representation toward query-relevant evidence while adapting its token budget to the amount of information required. Its key principle is simple. Additional tokens are allocated until the visual features relevant to the query are represented with sufficient fidelity. Starting from frozen encoder features, QART recurrently constructs a compact representation and uses reconstruction quality over the query-relevant visual support to control token allocation. On Visual7W and VQAv2, QART achieves 81.96% and 76.37% accuracy using only about 29 visual tokens, reducing token counts by 20-25x relative to dense VLM inputs. It further reaches 92.0% P@1 on CORE and 36.3% PR-AUC on PODS, outperforming SteerViT on PODS by 8.4%. Together, these results show that visual representations need not be fixed in either content or capacity and can instead adapt to what each query requires.

0.18

Pronouncing Hard Words: Phoneme Adaptation of a Pretrained Conversational Text to Speech Model

cs.CL eess.AS ·2026-10-16 ·Yashish M. Siriwardena, Dinesh K. Gnanasekaran, Tanmay Laud, Shinji Watanabe, Subhabrata Mukherjee +1 more
Abstract · 5 pages

Voice agents built as cascaded automatic speech recognition (ASR)->large language model (LLM)-> text-to-speech (TTS) pipelines increasingly rely on conversational TTS: streaming, context conditioned models that adapt delivery to dialogue history. Such models are natural and expressive but offer little explicit control over how rare, trust-critical drug names are pronounced, because these entities are too infrequent, and their vocabulary too fast-growing, to be covered by any amount of pre-, mid-, or post-training. We retrofit phoneme controllability onto a pretrained conversational TTS model without any pre/mid-training, using an ARPAbet vocabulary exten-sion, continued phoneme-augmented supervised fine-tuning (SFT), and a training-free inference-time acoustic-biasing step. On a 7,500 clip clinical drug-pronunciation evaluation, our model raises strict Gemini-judged correctness from 54.6% (no phoneme augmentation) to 90.7%, outperforming every conversational and stateless TTS baseline on the automatic metric. Because no training stage sees a drug name, the entire evaluation is out-of-domain, so the gains reflect learned pronunciation rather than memorization. In a blinded, randomized human study, adult listeners judge 84% of our clips correct, versus 82% for the strongest baseline, with overlapping confidence intervals on a stratified subset. Across all 250 matched clips, Gemini agrees with the human majority on 84.8% of binary labels (k{appa}=0.57).

0.18

Too Emotional to Be Fake? Intra-Utterance Emotion Shifts Expose Failure Modes in Speech Deepfake Detectors

cs.SD cs.CL ·2026-10-16 ·Jingyi Chen, Surya Koppisetti
benchmark
Abstract · Preprint. Under review

Expressive speech synthesis can now change emotion within a single utterance, yet its impact on deepfake detection remains unclear. We introduce EmoShift, a benchmark comparing emotion-stable (AA) and emotion-changing (AB) speech to examine these effects while accounting for static emotion identity and acoustic confounds. It comprises 7,240 bona fide utterances from four source datasets and 9,720 synthetic utterances generated using four expressive TTS systems. Across seven detectors, we find that emotion changes can make synthetic speech appear more genuine, increasing spoof miss rates by up to 25.3 percentage points on OMG-Emotion at fixed thresholds. Conversely, emotion-changing bona fide speech increases false-alarm rates by up to 24.4 percentage points on IEMOCAP. Further analyses reveal detector-specific sensitivities to emotion trajectories, not fully explained by static emotion identity or splicing artifacts. These findings underscore the importance of emotion dynamics in deepfake detection.

0.18

Hierarchical and Query-Aware Visual Token Pruning for High-Resolution LVLMs

cs.CV ·2026-10-16 ·Luther Wei, Ling Zou, Kang-Yang Huang, Yu-Syuan Xu, Wen-Huang Cheng
Abstract · 6 pages, 2027 ICCE

High-resolution Large Vision-Language Models (LVLMs) preserve critical, fine-grained visual details by encoding a global thumbnail alongside multiple local crops. However, the resulting thousands of vision tokens introduce severe computational bottlenecks during Large Language Model (LLM) inference. Existing token pruning methods rely primarily on query-agnostic visual saliency, while current query-aware approaches either operate at a coarse crop-level resolution or compute token relevance too late, after vision tokens have already entered the heavy LLM layers. To address this, we propose a training-free, hierarchical token pruning framework that utilizes query guidance prior to executing any LLM blocks. Our top-down hierarchy first operates at the instance-level, dynamically scaling the vision token retention ratio of each sample based on its visual complexity. Next, at the crop-level, we distribute this sample budget across local crops using sentence-level cross-modal similarity. Finally, at the token-level, we exploit projected vision tokens alongside textual query token embeddings in the shared space to measure fine-grained alignment, applying a text-debiasing normalization to produce high-fidelity query scores. These query-guided budgets and scores are fused with local and global visual saliency metrics to select the most critical vision tokens. Extensive experiments on LLaVA-NeXT-7B demonstrate that our plug-and-play framework retains 98.0% and 94.8% of unpruned performance at 25% and 10% vision token retention, respectively. Notably, at 25% retention, our method delivers a 2.0x inference throughput speedup and even surpasses the unpruned baseline on POPE by 0.7 points.

0.17

FederatedBytes: Tokenizer-Free, Byte-Level Small Language Models for Efficient Multimodal Federated Learning

cs.CL cs.DCcs.LG ·2026-10-16 ·Ali Sarabadani, Massimo Piccardi, Mehran Abolhasan, Farzad Tofigh, Ian Zhou +1 more
Abstract

Federated Learning (FL) is a powerful machine learning framework that allows building a global model from the contributions of multiple, distributed clients, while ensuring the privacy of their local data. However, its heavy computational and communication overheads challenge its use for contemporary AI workloads such as natural language processing (NLP) and multimodal analysis. For this reason, this paper presents a novel small language model that leverages byte-level text and multimodal data processing, enhanced by a linear attention mechanism, to enable manageable Federated Learning of NLP and multimodal analysis tasks. The proposed model, nicknamed FederatedBytes, manages to process text and any other data types, including images, by treating its input as a mere sequence of bytes, while its lean linear attention mechanism allows abating the overall memory and computational footprint. To prove the effectiveness of the proposed model we have carried out an extensive performance evaluation on sentiment analysis tasks in multilingual and multimodal settings over the Multilingual Amazon Reviews and ScienceQA datasets. The experimental results show that FederatedBytes has been able to outperform comparable models such as ByT5 and LSTM, achieving higher accuracy while significantly reducing memory and communication overheads.

0.17

Enhancing RAG-based Medical VQA with Gated Multimodal Representations

cs.CV cs.CL ·2026-10-16 ·Daniela Jimenez-Pano, Hugo Jair Escalante, Delia Irazú Hernández-Farías
benchmark
Abstract · Preprint submitted to ECIR

Medical Vision-Language Models (Med-LVLMs) have shown strong performance in Medical Visual Question Answering (MVQA), yet they may still produce factually inconsistent responses. Retrieval-Augmented Generation (RAG) can mitigate this issue by incorporating external clinical evidence; however, retrieval is often guided by a single modality, potentially overlooking the joint semantics of medical images and clinical questions. We introduce a multimodal RAG framework that learns a joint query representation through a Gated Multimodal Unit, enabling evidence retrieval to leverage information from both modalities while keeping the underlying Med-LVLM frozen. We evaluate the proposed approach on benchmark datasets. Results show that the multimodal query representation consistently outperforms a reference Med-LVLM and remains competitive with state-of-the-art methods. Our findings demonstrate the potential of multimodal query representations for more effective evidence retrieval in Medical VQA.

0.17

Interpreting SSL Representations for Spoof Detection: a WavLM Study

cs.SD cs.LG ·2026-10-16 ·Mohamed Mallat, Michele Panariello, Massimiliano Todisco, Nicholas Evans, Anthony Larcher
Abstract · 5 pages, 4 figures, 1 table. Published at IWBF 2026. An extended abstract of this work was presented as a poster at Odyssey 2026

Self-supervised speech models such as WavLM achieve strong spoofing detection performance, yet which layers encode the most spoof-relevant information and what acoustic properties they capture remain opaque. We present a layer-wise interpretability study of WavLM using the ASVspoof 5 database, combining three complementary analyses: (i) per-layer linear probing with attack-level evaluation across speech synthesis, voice conversion, and adversarial attacks; (ii) a canonical correlation analysis (CCA) linking layer representations to spoofing-relevant acoustic features; (iii) an exhaustive grid search over layer subsets showing that use of five layers outperforms full thirteen-layer pooling. CCA analysis reveals that, while most acoustic features peak in alignment with mid-layers and then decay, voice quality measures jitter, shimmer, and the harmonics-to-noise ratio show the opposite trend, with correlation increasing toward later layers. Embedding visualisations reveal that deeper layers capture a continuous acoustic similarity among attacks rather than discrete categories, suggesting these layers encode fundamental synthesis artifacts rather than attack-specific signatures.

0.17

BLANC: Adaptive Temporal Scheduling for Sign Language Production

cs.CV cs.AI ·2026-10-16 ·Youngmin Kim, Seoyeong Hwang, Sujung Hong, Seong Jae Hwang
code/data
Abstract · Project page: https://winston1214.github.io/BLANC/

Sign language production (SLP) aims to synthesize expressive signing motion from written text. Modeling signing motion presents two key temporal challenges. First, fine-grained hand articulations demand higher temporal resolution than relatively slower body movements. Second, signing events unfold over variable durations, calling for adaptive temporal scheduling during generation. To address these challenges, we first introduce a Multi-Rate Vector Quantizer (MRVQ), an articulation-decoupled motion tokenizer that represents the body and both hands as separate discrete streams at distinct temporal resolutions. Building on this representation, we propose BLANC (BLock-wise Adaptive diffusioN with Articulatory Clock), a gloss-free SLP framework that generates variable-duration motion blocks causally while denoising tokens in parallel within each block. At its core, a Latent Articulatory Clock (LAC) learns temporal structure from multi-scale motion-derived partitions to predict block durations and sequence termination during generation. We further introduce inference-time body-conditioned hand refinement to improve coordination across articulators. Extensive experiments on PHOENIX-2014T and CSL-Daily demonstrate that BLANC achieves state-of-the-art performance in sign language production.

0.17

ReVision: Reprogramming Image Generators for Dense Visual Perception

cs.CV ·2026-10-16 ·Yitong Yang, Yinglin Wang, Xuexin Liu, Wei Chen, Zhaoyang Xu +4 more
Abstract

Recent image generators exhibit rich visual priors that can support dense visual perception. However, existing approaches typically encode multiple perception tasks into a shared parameterization, making joint training sensitive to the mixture and weighting of heterogeneous data and prone to cross-task interference. To address this issue, we introduce ReVision, an instruction-conditioned parameter generation framework built on a frozen multimodal understanding model and a frozen image generator. Given an image and a free-form instruction, the understanding model jointly encodes them into a multimodal image--instruction representation, which a trainable parameter generator maps to transient low-rank updates for the generator. Each image--instruction pair induces a distinct set of transient prediction parameters, enabling instance-specific adaptation while both pretrained backbones remain unchanged. ReVision requires no task-specific heads, explicit task identifiers, or stored adapters. A shared RGB output interface unifies segmentation, depth estimation, and surface-normal prediction. Across diverse dense prediction benchmarks, ReVision achieves competitive performance with state-of-the-art methods while supporting flexible task specification through natural-language instructions. Our analysis further reveals substantial gradient disagreement across tasks, highlighting the conflict introduced when heterogeneous objectives share a single parameterization. Meanwhile, the generated updates exhibit structured task- and input-dependent organization in weight space, further validating instance-level parameter generation for multi-task perception.

0.17

From Adult Renal Masses to Pediatric Wilms Tumors: Evaluating Cross-Domain Generalization of Kidney Lesion Segmentation on MRI

cs.CV ·2026-10-16 ·Sarah de Boer, Myrthe A. D. Buser, Hartmut Häntze, Jawed Nawabi, Lisa C. Adams +6 more
Abstract

Deep learning segmentation models for pediatric oncology are often trained on small, single-center cohorts, whereas much larger datasets exist for adult diseases, such as kidney cancer. Whether models trained on adult renal lesions generalize to pediatric Wilms tumors remains unclear. We trained a segmentation model on single-sequence T1- or T2-weighted adult kidney MRI and evaluated it on a held-out adult test set and a pediatric Wilms tumor cohort, where it was compared with a pediatric-specific nnU-Net. Segmentation performance was quantified using the Dice similarity coefficient (DSC) and 95th-percentile Hausdorff distance. On the adult test set, the model achieved a median DSC of 0.86 for renal lesions. On the pediatric cohort, the adult-trained model achieved a median DSC of 0.85 compared with 0.90 for the pediatric-specific model. Despite differences in age, anatomy, and tumor type, the adult-trained model showed promising performance on pediatric Wilms tumors, with only a modest difference in median DSC compared with the pediatric-specific model. However, non-inferiority could not be shown, highlighting the value of pediatric training data. In future work, a fine-tuning strategy could be explored.

0.17

TransNSNet: A Transformer-Enhanced Multi-Scale Adversarial Framework for CT-to-PET Image Synthesis

eess.IV cs.CV ·2026-10-16 ·Shaojun Li, Xiaoyu Deng, Ikuo Keshi
Abstract · 21 pages, 7 figures. Supplementary material included. Under review at Biomedical Signal Processing and Control

This study investigates CT-to-PET image synthesis using TransNSNet, a Transformer-enhanced multi-scale adversarial framework. A paired CT-PET dataset was curated from the public non-small cell lung cancer (NSCLC) Radiogenomics collection through patient screening, slice-level matching, preprocessing, and patient-level train-test separation. The proposed framework generates PET-like predictions at multiple spatial scales and applies coarse-to-fine reconstruction supervision, while TransPatchGAN introduces contextual feature interaction into adversarial evaluation. A progressive ablation study compares Pix2Pix, NSNet-Base, NSNet-D, and TransNSNet. Image-level performance was summarized over a fixed convergence window (epochs 80-101) from one training run per model. TransNSNet achieved the best mean value in nine of ten metrics, whereas NSNet-Base achieved the highest visual saliency-induced index. For fixed-fraction high-intensity regions, NSNet-D was marginally higher in slice-level recall and balanced accuracy, but the equal-weight patient-level values were nearly identical and the highest Dice values shifted to TransNSNet. These differences were small relative to between-patient variability. In the separate adaptive patient-level analysis (n=10), no pairwise model difference remained statistically significant after Holm correction. TransNSNet also had a substantially smaller parameter footprint than the original next-scale models. These findings indicate promising image-quality trends, but they do not establish statistical superiority or clinical validity; synthesized PET-like images should not be interpreted as substitutes for clinically acquired PET.

0.17

MaskMerge: Structure-aware Token Merging for Efficient Instance Segmentation

cs.CV ·2026-10-16 ·Bavesh Balaji, Letian Jiang, Alexander Wong, David Clausi, Sirisha Rambhatla
Abstract

Token merging improves the efficiency of Vision Transformers by combining redundant tokens, but existing methods largely rely on feature similarity to determine which tokens to merge. This criterion is effective for semantic segmentation, yet can fail in instance segmentation, where adjacent objects of the same class often have highly similar features. We identify this failure mode as cross-instance token fusion. We introduce MaskMerge, an instance-aware token merging method that uses coarse intermediate mask predictions to constrain feature-based merging and preserve instance boundaries. We evaluate MaskMerge on two diverse instance segmentation domains characterized by high inter-instance visual similarity: PanNuke and iSAID. Against the SOTA token merging method ALGM, MaskMerge improves mAP from 30.1 to 32.1 on PanNuke while reducing computation from 229 to 153 GFLOPs and running 1.18x faster, and from 33.1 to 36.2 on iSAID while reducing computation from 198 to 182 GFLOPs and running 1.23x faster. Additionally, we also test it on Cityscapes and match ALGM's performance while using fewer FLOPs. The gains also transfer to an MAE pretrained backbone, demonstrating that instance-aware merging enables more aggressive compression when appearance similarity alone cannot reliably preserve object identity.

0.17

Knowledge-Guided Distributional Prompt Learning with Visual Hierarchy for Zero-Shot Anomaly Detection

cs.CV ·2026-10-16 ·Jiyul Ham, Seongho Yoon, Yonggon Jung, Sangjae Lee, Jun-Geol Baek
Abstract · 22 pages, 12 figures

Zero-shot anomaly detection (ZSAD) aims to detect anomalies in previously unseen classes using auxiliary images, with applications such as industrial visual inspection and medical imaging. Recent vision-language-model-based approaches improve ZSAD by aligning textual and visual representations, while distributional prompting further models prompt variability for unseen categories. However, existing methods do not explicitly combine normality-referenced distributional modeling with hierarchy-specific prompt specialization, limiting their ability to model anomalous semantics relative to normality while addressing both image-level detection and fine-grained localization. We propose Hierarchical Knowledge-Guided Distributional Prompt Learning (HiDist), a normality-referenced hierarchical distributional prompt framework for ZSAD. HiDist models each prompt token with a diagonal Gaussian distribution and generates stochastic prompt variations through Monte Carlo sampling. Rather than learning normal and anomalous distributions independently, anomalous prompt distributions are parameterized as deviations from learned normal references, explicitly encoding their dependency. The resulting distributions are specialized into global and local representations for image-level detection and pixel-level localization, respectively, while an adapted global visual representation guides hierarchical alignment. Experiments on seven industrial and medical datasets demonstrate strong cross-dataset performance across anomaly detection and localization tasks. Extensive ablation and representation analyses further support the effectiveness of the proposed distributional modeling and hierarchy-specific representation learning.

0.17

X-MDD: LLM-Based Mispronunciation Detection and Diagnosis for Unseen Languages with Canonical References

cs.SD ·2026-10-16 ·Xintong Wang, Junchuan Zhao, Ye Wang
Abstract

Mispronunciation detection and diagnosis (MDD) provides phone-level feedback to language learners, but extending it to new languages typically requires collecting and annotating learner speech for each target language. This repeated annotation effort is costly and limits the language coverage of pronunciation feedback. We study whether MDD capabilities learned from annotated learner speech can be applied to languages unseen during MDD training when canonical pronunciation references are available. Our multilingual LLM-based model, X-MDD, uses canonical speech and phones as references to predict the phone sequence actually produced by a learner. It incorporates in-context demonstrations for inference-time adaptation and is further optimized with group relative policy optimization (GRPO) using detection F1 and phone error rate. We further analyze how canonical inputs affect prediction and find that (1) canonical speech and phones jointly contribute to actual-phone prediction, and (2) conflicting speech evidence reduces the tendency to copy canonical phones. Trained only on Mandarin and English, X-MDD achieves a detection F1 of 38.07 on Arabic Qur'anic recitation, compared with 44.14 for a supervised challenge baseline trained on Arabic. It also achieves F1 scores of 78.68 and 58.78 on Mandarin and English, respectively, demonstrating effective multilingual MDD within a single model.

0.17

EGGHead: Efficient Generalizable Gaussian Heads from a Single Image

cs.CV ·2026-10-16 ·Pilseo Park, Andrew Hou, Michel Sarkis, Ning Bi, Yiying Tong +1 more
Abstract

Creating photorealistic human heads typically involves trade-offs between representation power, efficiency, and generalizability. Traditional mesh-based solutions relying on 3DMMs generally sacrifice representation power but are efficient to render. On the other hand, NeRF-based methods improve representation power and fidelity while sacrificing efficiency. Gaussian splatting methods attempt to resolve these trade-offs by offering both high representation power and efficiency with real-time rendering. However, their efficiency nonetheless depends on the number of rasterized Gaussians, and the time needed to optimize and prepare the head (the enrollment time) is often high. In light of this, we propose EGGHead, which introduces a novel Gaussian allocation algorithm that efficiently assigns Gaussians to the head's semantic regions given a fixed Gaussian budget. Our algorithm reduces the number of Gaussians needed while minimizing the impact on representation power. EGGHead operates in a single forward pass and outperforms previous single-image face modeling methods in reconstruction quality, novel view synthesis, and rendering speeds.

0.16

MLLM Frame Subset Ensembling for Audio-Visual Video QA and MLLM-based Reranking for Ad-hoc Video Search in TRECVID 2025

cs.CV ·2026-10-16 ·Andreas Goulas, Damianos Galanopoulos, Ioannis Patras, Vasileios Mezaris
survey
Abstract · Proc. TREC 2025 Workshop, Dec. 2025

This paper presents the overview of the runs related to the Ad-hoc Video Search (AVS) and Video Question Answering (VQA) tracks of TRECVID 2025 on behalf of the CERTH-ITI team. For the AVS track, we introduce a two-stage framework built on foundation models. In the first stage, multiple vision-language models (VLMs) encode both the input query, augmented through LLM-generated rephrasings, and the candidate video shots, producing weighted similarity scores for initial retrieval. In the second stage, we utilize a Multimodal-LLM(MLLM)-based reranking module that evaluates the semantic alignment between each shot among the top-N highest-ranked ones and the original query, generating updated relevance scores for reordering these shots. This MLLM-driven reranking significantly improves contextual matching and produces more accurate final rankings without requiring any model training. Regarding the VQA track, we fine-tune an audio-visual MLLM model on the provided TRECVID training dataset and we implement an inference-time scaling technique to enhance the multimodal understanding capabilities of the MLLM. For the open-ended Answer Generation (AG) task, we aggregate multiple model responses per question via a majority vote. The responses are generated with greedy sampling from different random frame subsets of the video and they are ranked based on the number of votes. For the Multiple-Choice (MC) task, instead of voting, we use mean pooling on the logits assigned by the fine-tuned model to each candidate response. Through the combination of fine-tuning and frame subset ensembling we achieve the highest score across 3 metrics in the VQA AG task and the second highest in the VQA MC task.

0.16

PanoBench: A Panoptic Occupancy Prediction Benchmark for Autonomous Driving

cs.CV ·2026-10-16 ·Yi Feng, Zizhan Guo, Junwu E, Yu Ma, Mengtan Zhang +6 more
benchmark
Abstract

Panoptic occupancy unifies 3D geometry, semantics, and instance identities, but its development is limited by scarce instance-aware annotations. Existing benchmarks mainly provide semantic labels reconstructed from incomplete sensor observations, while panoptic datasets have limited domain coverage or annotation completeness. We introduce PanoBench, a camera-centric panoptic occupancy prediction benchmark spanning synthetic and real-world driving scenes. It combines PanoBench-CarlaOcc, PanoBench-nuScenes, and PanoBench-KITTI360 through canonical data interfaces and unified evaluation protocols, providing multi-resolution annotations with persistent instance identities. The synthetic branch derives complete panoptic geometry from simulated meshes. For real-world data, a new pipeline reconstructs backgrounds using geometric priors and topology constraints, generates foreground mesh assets from multi-frame multi-view images, and fuses them into sparse panoptic voxels. A unified valid-mask protocol evaluates complete thing instances supported by temporal observations while scoring stuff regions at the voxel level. Analyses of observation fidelity, geometric completeness, spatial continuity, and temporal consistency demonstrate the benefits of the resulting annotations over existing occupancy benchmarks. We benchmark representative methods on semantic occupancy prediction, panoptic occupancy prediction, and 4D occupancy forecasting, and study voxel resolution, synthetic-to-real transfer, and storage efficiency. These results reveal complementary strengths of volumetric and ray-based metrics and persistent challenges in real-world instance-aware prediction. PanoBench is available at https://mias.group/PanoBench.

0.16

ODSS-Net: Orthogonally Decoupled Shared-Specific Representations for Multi-Modal Object Re-Identification

cs.CV ·2026-10-16 ·Zhifeng Ma, Yuhong Feng, Chongyi Huang, Mingyang Zhou, Minwen Deng
Abstract · ICASSP 2027 conference submission

Multi-modal object re-identification (ReID) leverages complementary information from heterogeneous sensors. Cross-modal alignment and fusion remain challenging. Existing methods mainly reduce modality gaps via feature alignment and fusion. However, alignment may entangle modality-shared structural information with modality-specific cues, while conventional fusion may not fully exploit complementary information. To address these issues, we propose ODSS-Net, a framework for orthogonally decoupled shared-specific representation learning and adaptive fusion of complementary modality-specific features. First, a Shared Structural Feature Construction (SSFC) module suppresses modality-dependent appearance variations and extracts modality-shared structural information. Then, an Orthogonal Specific Feature Learning (OSFL) module aggregates identity-relevant local information and removes the shared-aligned component via geometric orthogonal projection. Finally, a Sample Adaptive Mixture of Experts (SAMoE) module uses the shared structure as routing context to adaptively refine and fuse modality-specific features. Experiments on three benchmarks demonstrate the effectiveness of ODSS-Net for multi-modal object ReID.

0.16

DriftTTS: One-stage Single Step Text-to-Speech via Generative Drifting

cs.SD ·2026-10-16 ·Jaehyun Ahn, Dohyun Kim, Heeseung Kim
code/data
Abstract · Project page: https://drifttts.github.io/

Text-to-speech models built on score matching and flow matching produce natural speech, but only by calling the decoder many times, which makes synthesis slow. Existing ways to cut the number of calls either distill from a multi-step teacher or add a stage on top of a pretrained model, leaving single-step TTS without a pretrained TTS model unexplored. We question whether pretraining a TTS model is necessary, and propose DriftTTS, which adopts generative drifting, an objective that moves iterative sampling from inference into training. Drifting assumes samples of fixed shape, whereas speech has variable length; we address this issue with segment drifting. On LJSpeech, DriftTTS matches the multi-step quality of Matcha-TTS and RapFlow-TTS in a single decoder call, while being faster. To our knowledge, this is the first application of generative drifting to text-to-speech.

0.16

A Paired RGB-Sonar Dataset for Fish Monitoring with Consumer Sonar

cs.CV ·2026-10-16 ·Mourad Derbedrosian, Michael Hobley, Pietro Perona
Abstract

Imaging sonar enables underwater observation in conditions where optical imaging is ineffective due to darkness and turbid water. However, imaging-sonar systems previously used for ecological monitoring have typically been expensive, limiting their accessibility. Consumer forward-looking sonar has the potential to make fish monitoring more accessible, but its noisy imagery makes annotation and the development of automated methods challenging. We introduce the Huntington and Caltech Koi RGB and Sonar dataset (HaCKRS), which pairs Garmin LiveScope imaging-sonar recordings with top-down RGB video of koi carp. The dataset contains more than 105,000 temporally matched frame pairs, spanning approximately 3.3 hours of recordings across four range and orientation configurations. We provide frame-wise metadata, sonar geometry and coordinate information, synchronization data for all four configurations, and planar registration for the three horizontal configurations. The release focuses on paired observations and calibration rather than task-specific fish annotations. We intend HaCKRS to support annotation and the development of fish detection, tracking, and counting methods, and investigations into the potential of consumer sonar for ecological monitoring.

0.16

Look, Reason, Draw a Region: Recursive Refinement for Spatial Context Prediction

cs.CV cs.LG ·2026-10-16 ·Wajeeha Tahir, Anna Tegon, Lionnus Kesting, Gamze İslamoğlu, Kristóf Karacs +2 more
benchmark
Abstract · 17 pages, 6 figures, 3 tables

Spatial context prediction requires visual grounding and spatial reasoning to identify vacant regions relative to objects in a scene. Spatial relationships and physical constraints remain challenging for large vision-language models (VLMs). Recursive reasoning has demonstrated strong performance on discrete and abstract reasoning tasks, motivating its use for improving spatial context prediction in natural images. Weight sharing enables higher accuracy with a small trainable parameter count. We introduce Spatial-TRM, a compact recursive model with frozen vision-language features, a convex region decoder, and just 11.2M trainable parameters. This weight-tied reasoner supports six spatial relationships, three reference frames, and open-vocabulary anchor objects. On the RoboSpatial-Home benchmark, Spatial-TRM achieves 29.51% context accuracy, a gain of 9.02 percentage points (44.0% relative) over a separately trained single-step model with the same parameter count. Including the frozen encoders, the model has 877.8M total parameters, approximately $15\times$ fewer than the reported 13B size of RoboPoint. These results show that repeated refinement improves spatial context prediction from pretrained vision-language features at a fixed trainable parameter count on this benchmark.

0.16

Synthetic Image Oversampling for Rare-Class Representation Learning in Deep Facial Phenotyping

cs.CV ·2026-10-16 ·Miles Liao, Zhanliang Wang, Hongzhuo Chen, Kai Wang
Abstract · BIBM 2026 ML4BMI Workshop

Recent deep facial phenotyping methods can support the recognition of genetic syndromes, but patient images are often scarce and unevenly distributed across disorders. Synthetic facial images offer a way to supplement the training data without collecting additional patient photographs. Their usefulness, however, depends on whether they improve recognition of real patients: generated images may introduce artifacts or repeat features already represented in the training set. It is therefore necessary to examine both how much synthetic augmentation to use and which disorders benefit. We investigate disease-specific synthetic oversampling using images from the Phenotypic Disease Image Database to augment a subset of the GestaltMatcher Database. We compare increasing augmentation levels in an image-based model, GestaltMatcher-Arc, and a multimodal model, GestaltMML, while keeping the real training and patient-disjoint evaluation partitions fixed. Repeated training runs assess overall and disease-specific accuracy, and an exploratory embedding analysis examines changes in class compactness and separation. Augmentation improves mean recognition accuracy in both training setups, although gains remain modest in GestaltMatcher-Arc and continue across the tested levels in GestaltMML. These aggregate improvements are accompanied by uneven disease-level responses, including reduced accuracy for some disorders. In the embedding analysis, disorders with fewer real training images tend to gain more class separation, but geometric changes do not consistently track classification performance. Together, the findings support synthetic oversampling for limited-data facial phenotyping and emphasize the need to assess its amount and benefit for individual disorders within each training setup.

0.15

Beyond Clean Accuracy: The Practical Challenges and Benefits of Pixel-Based Language Models

cs.CL ·2026-10-16 ·Haowen Tang, Wei Xia, Raoyuan Zhao, Hinrich Schütze, Yihong Liu
Abstract · preprint

Pixel-based language models process rendered text without relying on a fixed subword vocabulary, offering a promising approach to noisy and multilingual inputs. However, their practical value depends on whether they can be reliably adapted to downstream tasks. To this end, we systematically examine this question through fine-tuning behavior, typographical robustness, and zero-shot crosslingual transfer. Under our tested settings, pixel-based models require longer training and broader labelled-data coverage than text-based models, while delayed learning and learning-rate-dependent failures make adaptation less reliable. Our analyses further suggest that task-relevant information is less readily accessible in pretrained pixel representations, placing greater demands on fine-tuning. Once successfully adapted, however, pixel-based models lose substantially less accuracy under typographical noise and achieve higher target-language accuracy than English-pretrained text-based counterparts on two crosslingual benchmarks, despite weaker clean English performance. Together, these findings reveal a central trade-off: replacing tokenization with rendering reduces sensitivity to corrupted and unfamiliar inputs, but shifts a central bottleneck to downstream adaptation.

0.15

Candidate-dependent residual routing trades top-of-list accuracy for recall in sequential recommendation

cs.IR cs.LG ·2026-10-16 ·Han Chen, Yingrui Li
Abstract · 33 pages, 3 figures. Includes 18 pages of Supplemental Information with 18 supplementary tables

Candidate-dependent routing lets a sequential recommender weight a user's past behaviors differently for each item it scores. We test whether this flexibility helps by adding a small residual routing module to a self-attentive baseline, in two versions that differ only in what sets the routing weights: the user's history representation or the candidate item's embedding. Across five paired seeds on the REES46 cosmetics-shop dataset, with matched training for all three models, candidate-dependent routing leaves the primary metric, normalized discounted cumulative gain at cutoff ten (NDCG@10), essentially unchanged (mean relative change +0.003%). Secondary and exploratory analyses show that this flat average hides two effects that cancel. The candidate-dependent model places more relevant items in its top ten (recall +1.81%) but ranks the first one lower (mean reciprocal rank -1.09%), and most of its new hits land at positions six to ten. Its gains come from queries with at least one target among the user's last 50 events, while queries whose targets all lie outside that window get worse. In this form, candidate-dependent routing trades accuracy at the top of the list for recall further down.

0.15

Encoder-Free Multimodal LLMs Learn Implicit Encoders

cs.CL cs.AIcs.CVcs.LG+1 ·2026-10-16 ·David Harwath, Desh Raj, Xavier Velez, Sandra Arcos-Holzinger, Willie Jeng +3 more
Abstract · Submitted to ICLR 2027

We empirically analyze two recently proposed encoder-free multimodal LLMs, Gemma-4 12B and Inkling-Small. Unlike previous multimodal LLMs, these models do not use modality-specific encoders to extract tokenized embeddings representing audio and image input; instead, they directly feed windows of samples, pixels, or spectral features to the Transformer language model. We analyze whether the early layers of these models learn to function similarly to dedicated audio and visual encoders. We find that these early layers do in fact act as ``implicit encoders,'' encoding audio and visual inputs in a way that is largely invariant to their cross-modal context. Furthermore, when performing multimodal reasoning tasks, we show that independently encoding the modalities throughout these early model layers achieves the same accuracy as jointly encoding them from the first layer onwards. Finally, we use layer-wise probing to analyze the salient information captured by the implicit encoders and find their behavior to be similar to several popular encoder models.

0.15

A100: A Hybrid Vision Foundation Model for Agricultural Plant Image Analysis

cs.CV ·2026-10-16 ·Mayur Pratim Das, Ajai Dubey, Aksshay Mathew P, Sana Afreen, Rishi Kumar Soni +3 more
benchmark
Abstract · 11 pages, 5 figures, 5 tables

The use of accurate vision models for crop monitoring, disease diagnosis, and plant phenotyping in agriculture is hindered by the lack of expertly annotated data, especially in developing countries. We propose A100, an agricultural vision foundation model that is pre-trained using self-supervised learning on 3 million unlabeled images from fields covering more than 100 crop species. A100 uses a hybrid CNN-Transformer backbone that fuses MBConv stages for local feature extraction with Transformer stages for global contextual representation. A100 uses pre-training with the integration of DINO self-distillation, iBOT masked patch reconstruction, and KoLeo uniformity regularization, resulting in features that are better than ImageNet-12K supervised representations for agricultural applications without any fine-tuning. By utilizing a frozen backbone, the accuracy for crop classification of A100 is 95.37%, which is +2.56% better than ImageNet-12K, and reduces the leaf count error by 10.4% on the GroMo25 phenotyping benchmark, proving that the same backbone can be used for classification and regression tasks. After full fine-tuning, the performance of A100 is 99.90% on 32-class crop identification and 99.02% on 44-class disease detection. On the difficult PlantDoc dataset, A100 outperforms all baselines, such as CvT-13 (64.89% against 63.56%). Remarkably, A100 obtains 98.01% accuracy on plant classification using just 1% of the labelled data (about 284 images per category), showing great label efficiency for use cases with limited labeling. Under an identical protocol, A100 also outperforms the publicly released agricultural foundation models FoMo4Wheat and LeafVision by more than 14 points at 1% labels. We plan to publicly release our pretrained model to allow agricultural professionals to adapt A100 for various plant analysis applications using limited annotations.

0.14

Not Every Mention Is a Claim: Rethinking Object Hallucination Detection in Vision-Language Models

cs.CV cs.CL ·2026-10-16 ·Hsin-Ling Hsu, Neeraja Kirtane, Kuan-Hao Huang
benchmark
Abstract

Object hallucination in vision-language models (VLMs) is usually measured by checking each object a caption names against the image. Most benchmarks and detectors alike treat every object mention as a claim that the object is present. This assumption fails when a caption names an object without asserting it, as in "a bench for people to sit on", which mentions people without claiming that any are in the image. We call such mentions non-assertive. Treating them as hallucinations inflates the error rates attributed to models that use such language. To separate them from real hallucinations, we introduce NAME-Bench (Non-Assertive Mention Evaluation Benchmark), 25,497 object mentions from captions that five VLMs generate for COCO images. Each mention is labelled concrete (asserted and present), hallucinated (asserted and absent), or non-assertive, and non-assertive mentions are further divided into four subtypes: depictive, purpose, negation and hedged. All the hallucination detectors that we evaluate flag non-assertive mentions as hallucinations, at rates between 14% and 80%. Adding non-assertive mentions to detector training as negatives largely corrects this on LLaVA-1.5. On the four newer VLMs, a comparable correction takes five to ten times more data, and the detectors then catch fewer real hallucinations. These findings show that current detectors conflate mentioning an object with asserting it, and that this conflation is harder to undo on the four newer VLMs than on LLaVA-1.5. Hallucination evaluation should therefore separate what a caption asserts from what it merely mentions.

0.14

Truth in Contrast: Calibrated Factual Evidence from Language Model Likelihoods

cs.CL cs.AI ·2026-10-16 ·Ponhvoan Srey, Jason Lucas, Pittawat Taveekitworachai, Junyoung Lee, Anh Tuan Luu
Abstract

Large Language Models (LLMs) encode substantial factual knowledge. Yet, estimating confidence in factual claims reliably remains challenging. We study likelihood-based factual confidence from the complementary perspectives of discrimination and calibration across multilingual factual statements. Specifically, we instantiate contrastive generative likelihoods as Contrastive Factual Evidence (CFE), which compares the likelihood of the same claim under competing factual and non-factual conditions, and derive its probabilistic counterpart, the Contrastive Factual Posterior (CFP). We show analytically that CFP calibration depends on the differential error between the two surrogate conditionals, allowing errors shared across them to cancel and yield tighter calibration bounds. Empirically, CFP substantially improves calibration while maintaining factual discrimination comparable to its strongest competitor. Nonetheless, discrimination is uniformly weak across methods in lower-resource languages, which simultaneously undermines the utility of the measured calibration. To this end, we interpret a likelihood geometry, within which an evidence-bearing direction is induced, along with an orthogonal evidence-invariant direction. In our investigation, we selectively align target-language claims with stronger English evidence only along the factual-evidence direction, improving discrimination and open-ended factual generation without target-language labels.

0.14

Deep-Learning Algorithm for Indeterminate Thyroid Nodules: Comparison and Combination with Molecular Testing

cs.CV q-bio.QMeess.IV ·2026-10-16 ·Jichen Yang, Jeffrey Ashton, Jikai Zhang, Maciej A. Mazurowski, Benjamin Wildman-Tobriner
Abstract · 16 pages, 7 figures, 8 tables (3 supplementary)

Thyroid nodules with indeterminate cytology (Bethesda III) carry an intermediate risk of malignancy, and molecular testing is widely used to decide between surgery and surveillance, but it is costly and not universally available. We investigated whether a convolutional neural network (CNN) applied to routine ultrasound images can stratify Bethesda III nodules, and how it compares and combines with molecular testing. We retrospectively identified 466 nodule records from 418 patients who underwent molecular testing (Afirma or ThyroSeq). A CNN pre-trained on determinate nodules was adapted to indeterminate nodules with soft labels and transfer learning. After removal of duplicate records, 119 nodules with surgical pathology were used for development and 47 were held out for testing. On the test set, the deep learning (DL) model achieved an AUC of 0.73 (95% CI 0.51-0.94), compared with 0.55 (0.38-0.71) for molecular testing (P = 0.15). At a cut-off inherited from the original algorithm, the DL model was more specific than molecular testing (0.95 vs 0.32; P < 0.001), with a higher positive predictive value (0.67 vs 0.21), a lower sensitivity (0.44 vs 0.78; P = 0.25) and a similar negative predictive value (0.88 vs 0.86). All six DL-positive test nodules were molecular-positive, and the DL output split the molecular-positive nodules into a high-risk (4/6 malignant) and a lower-risk group (3/27 malignant; P = 0.01). Surgical verification was strongly driven by the molecular result (99% of positive vs 19% of negative results), and molecular-test accuracy depended on the platform. A logistic regression of the DL output, nodule size and the molecular result dropped the DL term and did not improve on the DL output alone. Ultrasound DL complemented molecular testing for Bethesda III nodules, in particular by identifying molecular-positive nodules at lower risk of malignancy.

0.14

LonGaussian: Tracking Foundation Priors for Long-Context Feedforward Gaussian Reconstruction

cs.CV ·2026-10-16 ·Zhoufeng Xie, Guanjun Wu, Shuai Zhang, Xinggang Wang, Wenyu Liu
Abstract

Reconstructing high-quality, renderable 3D scenes from long videos is essential for robotic perception and embodied intelligence. Many feed-forward Gaussian reconstruction methods rapidly predict 3D scenes from a small set of images, but extending them to longer sequences presents two bottlenecks: jointly processing all views increases GPU memory requirements, while accumulating pixel-aligned predictions redundantly represents the same surfaces and increases storage and rendering costs. We present LonGaussian, a streaming feed-forward framework that reconstructs compact, globally renderable Gaussian scenes from long image sequences with known intrinsics and unknown camera poses. Built on a pretrained streaming 3D reconstruction backbone, LonGaussian decodes long-context geometric features into dense Gaussians. Within each chunk, multi-frame point tracks link repeated observations of the same surface and guide a learned merger to consolidate the corresponding Gaussian predictions. Rendered-depth consistency filtering further suppresses Gaussians from other chunks that incorrectly occlude locally reconstructed surfaces. On DL3DV and RealEstate10K benchmarks, LonGaussian improves novel-view rendering quality while using substantially fewer Gaussians than the evaluated feed-forward baselines.

0.14

What Shapes Brain Alignment of Large Language Models during Dialogue? Insights from Language Production and Comprehension

cs.CL cs.LGq-bio.NC ·2026-10-16 ·Yuko Nakagi, Yuya Chiba, Yasue Kishino
Abstract · 33 pages

Brain alignment of large language models (LLMs) offers a way to investigate representational similarities between LLMs and the human brain during language processing. While previous studies have mainly focused on passive language comprehension, little is known about such alignment during dialogue, in which speakers engage in both language production and comprehension. In this study, using fMRI data collected during dialogue, we examine brain alignment during language production and comprehension across LLM layers and brain regions. We assess its association with the intrinsic dimension (ID) of LLM representations and the contributions of several interpretable linguistic and communicative features. Our results reveal distinct patterns of brain alignment between production and comprehension, as well as strong associations with ID in brain regions associated with each processing mode. Furthermore, we show that the contributions of linguistic and communicative features also vary across processing modes, LLM layers, and brain regions.

0.14

STC-MGNet: Spatial-Temporal-Channel Attention-Enhanced Multimodal Graph Network for Pedestrian Intention Prediction

cs.CV ·2026-10-16 ·Yuanzhe Li, Meng Wang
Abstract

Accurate prediction of pedestrian crossing intentions plays a critical role in enhancing the safety of pedestrian-vehicle interactions in autonomous driving. However, effectively capturing subtle spatiotemporal patterns in pedestrian pose sequences and integrating complementary visual and ego-vehicle motion information remain challenging. This paper proposes a spatial-temporal-channel attention-enhanced multimodal graph network (STC-MGNet) for pedestrian crossing intention prediction. STC-MGNet introduces a spatial-temporal-channel (STC) attention-enhanced pose graph encoder to learn discriminative representations from pedestrian pose sequences. Three parallel attention branches selectively emphasize informative skeletal joints, observation frames, and feature channels, with their outputs integrated through learnable fusion weights. To leverage complementary visual and ego-vehicle motion information, a multi-stage cross-modal channel modulation mechanism is introduced to adaptively refine pose representations through modality-specific channel gating at successive graph encoding stages. Experimental results demonstrate that STC-MGNet achieves accuracy rates of 70%, 87%, and 88% on the JAADbeh, JAADall, and PIE datasets, respectively, outperforming several representative baseline methods. Ablation experiments further highlight the complementary contributions of visual and ego-vehicle motion information to prediction performance.

0.13

Explaining Chess and Go in Natural Language

cs.CL cs.AIcs.LG ·2026-10-16 ·Ori Press, Alexander Golubev
Abstract

Modern chess engines play far beyond human ability, often choosing moves whose underlying ideas are difficult for humans to understand. We propose an automatic metric for evaluating chess explanations, grounded in chess-engine evaluations. Evaluating state-of-the-art language models using this metric, we find that they perform surprisingly poorly, often relying on surface-level board observations rather than robust, transferable chess concepts. We then introduce BinProp, a training-free method that enables these same models to discover reusable chess concepts and use them to explain moves in natural language. BinProp improves explainability for every model tested, nearly tripling the average score compared with direct prompting. Beyond chess, the same method discovers useful concepts in Go, draughts, and shogi without additional tuning.

0.13

SIT-Eval: Situation-Based Interactive Turn-by-Turn Evaluation of Conversational Agents

cs.CL cs.AI ·2026-10-16 ·Ravil Mussabayev, Saveliy Zavertyaev, Dmitriy Parpulov, Andrey Savchenko
Abstract

The rapid proliferation of conversational AI agents necessitates evaluation methods that assess performance across diverse dialogue scenarios, utilizing multiple situational criteria, as appropriate behavior depends on context and encompasses various aspects. Existing dialogue benchmarks typically use static natural conversations, which underrepresent rare situations---a critical safety gap, as low-probability events become frequent failures at production scale. However, limited sample sizes force a choice: either maintain natural dialogues for realism or turn to adaptive situation sampling for comprehensive yet statistically reliable coverage. We introduce SIT-Eval, an adjustable framework for evaluating situation-based conversational agents. We formalize an agent's desired behavior as situational rules: if situation is P, follow rules Q. At each conversation turn, an LLM-interlocutor enacts a uniformly sampled situation. A context-aware LLM-judge leverages all conditionally applicable rules as criteria to evaluate the agent's reply for that situation. The judge is validated using human-verified synthetic golden dialogues. Benchmarking 8 LLMs on 1292 conversation turns reveals failure modes hidden by aggregate metrics: models with similar overall scores vary widely in their adherence to individual rules. All models struggle most with ambiguity clarification and Socratic questioning.

0.12

Multi-author style detection with stylistic embeddings

cs.CL ·2026-10-16 ·Man-Ling Sung, Chak Fai Li, Damianos Karakos, Fred Bernardin, Brian Ulicny
Abstract · Submitted and accepted to Workshop on Natural User-generated Text (W-NUT) in Empirical Methods in Natural Language Processing (EMNLP), 2026

We investigate the use of Large Language Models (LLMs) in multi-author style change detection. Using off-the-shelf LLMs or supervised fine-tuning solely within the target domain fails to yield effective features; for this reason, we propose cross-domain transfer learning with LLM-based embeddings fine-tuned for authorship verification. We fine-tune Mistral-7B embedding model and compare it with DeBERTa, used in the state-of-the-art (SOTA) systems in PAN-25 multi-author writing style analysis challenge. Results show that DeBERTa and Mistral excel in different user-generated text scenarios. We exploit this diversity by combining both embeddings, leading to more effective style change detection.

0.12

Aligning Language Models Towards Counterfactual Invariance

cs.CL cs.AIcs.LG ·2026-10-16 ·Basar Demir, Emily Shen, Ksheeraj Sai Vepuri, Sourabh Kulkarni, Harihar Subramanyam +3 more
Abstract

Small variations in prompt characteristics can meaningfully alter the semantics, tone, and contextual assumptions of large language model outputs. Such characteristics can involve group variables such as demographics or other sensitive latent attributes. Sensitivity to such input variables is undesirable when demographic information is causally irrelevant to the task, but might still be relevant for generation of semantically coherent responses. This makes the task of enforcing counterfactual invariance while preserving model utility difficult. To study this problem, we first show that standard model alignment recipes (such as reinforcement learning) increase differences in response quality across demographics, while improving overall response quality, ultimately amplifying what is known as counterfactual bias. To address this, we develop a policy gradient technique involving a group-level reward function that operates on matched pairs of counterfactual prompts. For completeness, we build a dataset of controlled counterfactual prompt pairs and design an evaluation that separately measures cases where demographic information is causally relevant to the answer. Finally, we validate our findings on several fairness benchmarks. We demonstrate that our methods reduce counterfactual differences while preserving overall response quality and that the models continue to adapt appropriately when such sensitive information is relevant.

0.12

The Cognitive Cost of LLM Writing: A Reading-Time Comparison with Human Text

cs.CL ·2026-10-16 ·Kamer Cekini, Elisabetta Biondi, Chiara Boldrini, Andrea Passarella, Marco Conti
Abstract

Although linguistic differences between human-written and Large Language Model (LLM)-generated texts have been extensively studied, their implications for the cognitive effort involved in reading remain less explored. In psycholinguistics, reading time is commonly used as an indirect measure of cognitive processing effort. Building on this perspective, we investigate whether human-written texts and their LLM-generated rewrites exhibit differences in predicted reading time that may reflect differences in processing demands for readers. To this end, we introduce RoBERTa-RT, a model trained on human eye-tracking data to predict word-level reading times. We evaluate the model on both internal and external test sets, construct a paired dataset of 496 human-written articles and their LLM-generated rewrites, and apply RoBERTa-RT to compare their predicted reading times. Our findings show that LLM-generated rewrites tend to be shorter and have slightly lower predicted total reading times, while exhibiting higher predicted reading times per word. These differences are associated with systematic linguistic changes, including longer and less frequent words, greater lexical diversity, and higher contextual predictability. From a cognitive perspective, our findings suggest that LLM-generated writing may redistribute reading demands: reducing the amount of text to process while increasing the predicted processing effort associated with individual words. This highlights the importance of considering not only the linguistic characteristics of machine-generated content, but also its potential implications for human cognitive processing.

0.12

Landmark-Free and Scale-Blind Multi-View Registration of Preoperative Liver Models to Intraoperative Surfaces

cs.CV cs.AIcs.HC ·2026-10-16 ·Jiaming Feng, Gerardo Loza Galindo, Shahid Farid, Sharib Ali
Abstract · 10 pages

Augmented-reality guidance in laparoscopic liver resection requires aligning preoperative 3D liver models from CT/MRI with intraoperative surfaces. Monocular reconstruction in most cases cannot guarantee scale, albeit rigid-motion registration requires scale as an input. Our proposed framework ``SimCompose'' estimates the full similarity transformation $\Sim$, including scale, without relying on landmarks and applies the same transformation across multiple views with known relative geometry. However, reliable detection of registration failures remains challenging in surgical settings, where ground-truth alignment is generally unavailable. For this, we propose a geometry-based selector that evaluates candidate transformations without requiring ground truth. Large disagreement among these transformations is used to flag potentially unreliable registrations. Finally, we propose to use a non-rigid refinement based on a low-dimensional deformation model from Neo-Hookean simulations in the Simulation Open Framework Architecture (SOFA), without additional network training. SimCompose is trained exclusively on synthetic partial views. Its encoder processes point clouds normalised to a unit root-mean-square radius, without access to metric scale, and filters each view before multi-view fusion. Evaluation on unseen synthetic data (300 single-view pairs per run; 100 three-view registrations) shows that multi-view integration and candidate selection increase the success rate under stringent $\Sim$ accuracy criteria from 38.4% for a single-view, single-method configuration to 77%. Existing learning-based baselines do not satisfy these criteria even when provided with the exact scale. On five in-vivo patients with 92 keyframes, the rigid registration stage reduces surface error more than three-fold relative to learning-based baselines with external scale information.

0.12

ShadowCast: On-the-Fly Context-Aware Shadow Augmentation for Remote Sensing Image Segmentation

cs.CV ·2026-10-16 ·Dongho Yoon, Yechan Kim, Jongmin Joo, JongHyun Park, Moongu Jeon
Abstract

Shadows cast by elevated objects alter the spectral appearance and obscure the boundaries of objects in remote sensing images, thereby degrading semantic segmentation performance. Most existing approaches handle shadows as degradations to be detected and removed, requiring additional costs from dedicated networks and shadow-specific annotations. We propose ShadowCast, which takes the opposite approach: it synthesizes shadows and injects them during training for remote sensing segmentation. It extracts masks of elevated objects, such as buildings and trees, directly from the segmentation labels and estimates an image-specific cast-shadow direction cue by probing the brightness around them. It then projects synthetic shadows along the inferred cast-shadow direction, requiring no learned network and no acquisition metadata. Experiments on three widely used remote sensing benchmarks, Vaihingen, Potsdam, and LoveDA, show that ShadowCast consistently outperforms the baseline across diverse state-of-the-art segmentation architectures. Moreover, it brings additional performance gains when combined with standard augmentation and regularization techniques, indicating that stochastic shadow injection serves as a simple and complementary augmentation strategy for remote sensing segmentation.

0.11

Frame-Consistent Feed-Forward 3D Gaussian Splatting from Sparse Unposed Images

cs.CV ·2026-10-16 ·Viruthshaan Vinasirajan, Ranga Rodrigo
Abstract

Feed-forward Gaussian models have made novel view synthesis almost instantaneous. Feed-forward methods predict Gaussians directly, either in two stages or in a single merged pass. Using two separately trained models is attractive because each can specialize in its own objective: one for accurate pose estimation and the other for high-quality reconstruction, rather than forcing a single jointly trained network to compromise between the two. A natural instantiation pairs a frozen pose estimator, VGGT, with a frozen reconstruction backbone, TokenGS. Prior work, such as VGGT-X and 3R-GS, also adopts a two-stage design, but attributes the resulting quality degradation to noisy pose estimates and introduces learned refinement. We show that this is not the primary source of the problem. Through per-camera analysis across 107 scenes from two datasets, DL3DV and MipNeRF360, using COLMAP poses as the reference, we find that the rotation error after alignment is small (median 1.6 degrees), whereas the translation scale is offset by a large, dataset-consistent factor. Controlled ablations over 140 DL3DV scenes show that the primary obstacles are two integration failures rather than pose error: a coordinate-frame inconsistency and a translation-scale mismatch, both of which can be corrected deterministically. With these fixes, a frozen VGGT driving a frozen TokenGS achieves 22.47 PSNR, within 0.51 dB of the COLMAP reference, with no additional training, when poses are estimated from all scene views. In the sparse setting, where reconstruction uses only four context images under the standard pose-free protocol, the same pipeline matches this all-view quality, yielding a fully COLMAP-free pipeline without any refinement step.

0.11

From Spectrograms to Shallow Speech SSL: Audio-Frontend Design for Generator-Disjoint and Low-Latency Audio--Visual Deepfake Detection

cs.SD cs.AIeess.AS ·2026-10-16 ·Qiwu Wen, Jahangir Alam, Eric Charton
Abstract

Video-mediated conversational systems rely on synchronized speech and facial video, making audio--visual input integrity important for reliable perception. We study audio-frontend design in a fixed MRDF (Modality-Regularization-based DeepFake)-based pipeline with AV-HuBERT visual processing and an MSCT (multi-scale cross-modal transformer) fusion encoder. We compare spectrogram-style log-filterbank features with frozen wav2vec~2.0 and WavLM representations, dual-SSL fusion, executed depth, and shallow-layer aggregation. Across three training runs, full wav2vec~2.0 improves mean macro AUROC from $84.19\%$ to $87.19\%$ on generator-disjoint MMDF. Six-layer execution further improves detection, and learned non-uniform aggregation adds $2.53$ percentage points, reaching $90.61\%$. Shallow execution reduces measured inference latency relative to full-depth wav2vec~2.0 across the tested input durations; first-second evaluation reaches $86.85\%$ macro AUROC. In a single-run comparison, dual-SSL concatenation outperforms either single stream, while one-way cross-attention yields lower detection performance and higher latency than concatenation. Layer-weight analysis shows greater emphasis on early speech layers, while removing the selector's visual input has little effect on detection performance.

0.11

PromptSep-Speaker: Generative Target Speaker Extraction via Multimodal Prompting and Semantic Alignment

cs.SD eess.AS ·2026-10-16 ·Yutong Wen, Ke Chen, Prem Seetharaman, Haonan Chen, Oriol Nieto +3 more
Abstract · Submitted to ICASSP 2027

Target speaker extraction (TSE) isolates a desired speaker from a multi-speaker mixture using cues that identify the target. Prior work has explored acoustic and textual cues, such as enrollment audio and different types of speaker descriptors, but typically relies on a single cue family. Yet no single cue is both broadly available and reliably discriminative. Moreover, most TSE systems adopt mask-prediction models, which are prone to audible artifacts and generalize poorly across acoustic conditions. To address these limitations, we propose PromptSep-Speaker, a generative TSE model based on latent diffusion that unifies four interchangeable cues: speaker temporal order, transcript, enrollment audio, and speaker caption. The generative formulation delivers higher separation quality and multi-modal prompting improves applicability across scenarios. To fully exploit each cue, we further introduce a frozen WavLM representation with an auxiliary alignment loss that supervises query-mixture cross-attention without extra labeling. Across four evaluation sets and a listening study, PromptSep-Speaker achieves state-of-the-art performance over TSE and universal source separation baselines, improving separation accuracy and quality across cue modalities.

0.11

AcousticAgent: A Floorplan-Centric Simulation Platform for Distributed Multi-Room Speech Processing

eess.AS ·2026-10-16 ·Zhiqi Ai, Han Cheng, Shiyi Mu, Yongjin Zhou, Shugong Xu
dataset
Abstract

Existing acoustic simulation tools either generate room impulse responses (RIRs) efficiently in parameterized rooms or rely on preconstructed three-dimensional scenes, making it difficult to build large numbers of editable residential multi-room environments. We present AcousticAgent, a floorplan-centric acoustic simulation platform that converts structured floorplans into editable, acoustically connected multi-room scenes. AcousticAgent jointly models room geometry and connectivity, furniture, materials, sound sources, and microphone configurations. A scene-oriented Acoustic Materials Database maps semantic elements to acoustic parameters, while a floorplan-aware hybrid geometrical-acoustics solver generates multichannel RIRs and reverberant audio. The platform also supports directional and moving sources, distributed microphones, and VLM-assisted scene construction from floorplan images and text. Using AcousticAgent, we construct FP-RIR, a dataset covering 15,376 residential floorplans with diverse room topologies, propagation conditions, and source--receiver configurations. We validate its acoustic fidelity against real-world measurements and evaluate whole-home source localization, speech enhancement, and far-field speech processing under single-device and distributed microphone settings. The results demonstrate the effectiveness of floorplan-aware modeling for large-scale residential acoustic simulation and the benefits of distributed spatial processing in multi-room environments. AcousticAgent and FP-RIR provide scalable and configurable resources for developing and evaluating residential speech systems.

0.11

BiNar: Bitstream Native Restorer for Bitstream-corrupted Image Restoration

cs.CV ·2026-10-16 ·Fangcheng Li, Zhen Yu, Kejun Wu, Tianyi Liu
Abstract

Bitstream-corrupted image restoration (BCIR) aims to reconstruct faithful images from corrupted compressed bitstreams. However, corruption occurs in bitstream-domain may lead to unpredictable error propagation and distortion amplification in pixel-domain, and even result in complete undecodability by standard decoders, posing significant challenge to existing image restoration paradigms. To address this challenge, we propose Bitstream Native Restorer (BiNar), a generative restoration framework that leverages fine-grained bitstream-native semantics to guide bitstream-corrupted image restoration. In BiNar, Bitstream Semantic Modeling (BSeM) extracts robust informative semantics and models sequential bytes directly from corrupted bitstream, enabling to generate fine-grained bitstream descriptions without relying on pixel decoding. Based on the descriptions,the Corruption-adaptive Multimodal Semantic Aggregation (CMSA) further extracts adaptive visual modal cues in decoded corrupted images and aggregates to bitstream descriptions. The multimodal aggregated fine-grained descriptions can serve as coherent restoration guidance. Finally, Blind Corruption Localization (BCoL) exploits the segmentation prior of visual foundation models to identify and locate error regions that requires recovery in corrupted images. The located error regions and the fine-grained descriptions jointly serve as visual prompt and language prompt of diffusion models image restoration, respectively. Extensive experiments under diverse bitstream corruption conditions show that BiNar outperforms all State-of-the-arts image restoration models regardless of perceptual quality or objective quality. Ablation studies demonstrate the effectiveness of fine-grained bitstream-native semantics in BCIR task.

0.11

Controlling the Diversity-Quality Trade-Off of LLMs by Merging Models Before and After Alignment

cs.CL cs.LG ·2026-10-16 ·Yasuaki Sumita, Tomoharu Iwata, Toshiyuki Tanaka
Abstract · 13 pages, 5 figures

Alignment is known to change the output characteristics of large language models (LLMs). Base models tend to generate more diverse outputs, whereas aligned models tend to produce higher-quality outputs. The appropriate balance between the two depends on the task. Existing methods either require costly fine-tuning or, like temperature sampling, mainly improve lexical diversity rather than semantic diversity. We introduce a depth-aware FrankenMerging approach for controlling the diversity--quality trade-off between an aligned model and its base model without additional training. This approach combines the lower layers of an aligned model with the higher layers of its base model based on the hypothesis that lexical diversity is more closely associated with lower layers and semantic diversity with higher layers. Experiments showed that varying the number of layers taken from each model adjusts both diversity and quality. In most cases, the merged models exhibited higher semantic diversity than the aligned model, and at some split positions, they showed higher lexical diversity than both original models. They also demonstrated a better diversity--quality trade-off than temperature sampling in most settings. A supporting ablation, in which we reversed the merging order, yielded results largely consistent with the hypothesis.

0.10

ULIFLOW: Unified Latent Human-Object Interaction Flow

cs.CV ·2026-10-16 ·Wenqi Cai, Chunzhi Gu, Weichen Zhou, Yawen Zou, Yijin Wei +3 more
position
Abstract

Text-conditioned 3D human--object interaction (HOI) generation remains challenging because individually plausible human and object motions do not necessarily form a coherent interaction. We argue that a key limitation lies in conventional skeleton-based representations, which compactly describe body kinematics while leaving surface-level interaction geometry largely implicit, often requiring additional contact-specific regularizers. We propose ULIFlow, a framework that explicitly models continuous human--object relations within a unified latent space. To this end, we derive continuous marker--object relations from sparse body-surface markers and treat them as an explicit interaction modality alongside human motion and object dynamics. A Mixture-of-Transformers (MoT) autoencoder is then designed to jointly model these heterogeneous streams while preserving their modality-specific structures. Importantly, we let the relation stream directly participate in human and object reconstruction, encouraging the shared latent representation to preserve both individual dynamics and the interaction geometry that coordinates them. To further achieve variable-length generation, we introduce a blockwise Flow Forcing scheme with global sequence-progress conditioning, enabling progressive continuation from fixed-length training windows while maintaining consistent long-horizon dynamics. Extensive experiments demonstrate that ULIFlow consistently outperforms prior work in overall generation quality and human--object coordination, while preserving coherent interactions over requested durations beyond the training horizon.

0.10

MUB: An Identity-Masked Benchmark for Music Understanding in Multimodal Language Models

cs.SD ·2026-10-16 ·Kun Liu, Lyra Tao
benchmark
Abstract

Evaluating musical perception in multimodal language models requires separating audio evidence from answers recoverable through metadata or external tools. We present the Music Understanding Benchmark (MUB), a locally assembled collection of 500 identity-masked recordings with existing annotations for tempo, key, time signature, and structural intervals. The collection covers all 24 major/minor key combinations, tempos of 47--180 beats per minute, and 4,676 annotated intervals. Original titles, artists, and platform identifiers are excluded from evaluation inputs. Relative to recent music question-answering and temporal-grounding work, MUB emphasizes scalar tempo prediction and full-recording interval outputs, together with a documented metadata-masking protocol. Saved baseline predictions yield 53.6\% tempo accuracy for a simple DSP estimator and 98.4\% time-signature accuracy for a constant 4/4 predictor. We identify important limitations, including an asymmetric legacy key metric, unverified independent annotation agreement, extreme metre imbalance, and unresolved recording-level rights clearance. No contemporary multimodal language model comparison has yet been completed. MUB is therefore an evaluation resource and protocol under validation, rather than a completed demonstration of model capabilities.

0.09

LiveSync: Iterative Self-Training for Lip-Sync Generation

cs.CV ·2026-10-16 ·Huaishao Luo, Huan Zhang, Guanbo Bao, Yinglu Liu, Youzheng Wu +1 more
Abstract

Audio-driven lip synchronization aims to match target speech through localized mouth and jaw articulation while preserving subject identity, head pose, and surrounding facial context. Naturally recorded videos provide synchronized reconstruction targets but rarely offer pairs that are matched in identity and pose yet differ primarily in speech-related articulation. Mask-based training enables reconstruction from these videos, but random-reference inputs differ from the aligned self-reference used at inference. We present LiveSync, an iterative self-training framework in which a single mask-based model generates pseudo pairs and is refined for deployment. The model is first trained on a curated real-video corpus augmented with externally generated clips covering under-represented visual conditions. A frozen checkpoint then uses independently sampled audio to generate mouth-edited references from real-video frames, copying pixels outside the edit mask from each source frame. After confidence filtering, these references are paired with aligned masked inputs and original real-frame targets. The retained pseudo pairs are mixed with real training data to refine the same model from its current weights, updating training supervision without adding inference stages. Although we instantiate the framework with a single-pass GAN-based model, the self-training strategy could in principle also be adapted to diffusion-based models. Experimental results show that LiveSync achieves leading synchronization performance on HDTF and competitive performance compared with the baselines on the AIGC-LipSync subset, while maintaining low inference latency.

0.07

Security & Privacy 33

ECHO: Replacing the Compliance Proxy with Response Harmfulness in Optimization-Based Jailbreak Attack

cs.CR cs.CL ·2026-10-16 ·Yuanwei Wu, Lingdong Huang, Yunlong Mao, Kanye Ye Wang, Jingyu Hua +1 more
top venuecode/data
Abstract

Despite extensive safety alignment, large language models remain vulnerable to jailbreak attacks. Existing optimization-based attacks share a single objective: maximize a token-level proxy of victim compliance---such as the likelihood of an affirmative prefix---computed on the prompt rather than the decoded reply. This proxy cannot distinguish a cautious hedge from a fully actionable harmful answer, leaving the search blind to what the model actually says. We propose ECHO (Evaluation-Conditioned Harmfulness Optimization), which replaces the compliance proxy with response-level, continuous harmfulness scores from HarmSig, a 1.7B regressor grounded in actual harmful capability. ECHO couples HarmSig with five word-level perturbation operators and Keyword-Anchored Beam Search with global top-m memory, under a strict black-box threat model. Across five open-weight victims on AdvBench, ECHO achieves up to 0.898 StrongREJECT mean score and 89.23% HarmBench success. Our code is publicly available at https://github.com/STANWYW/echo.

0.46

Provable Compliance: A Protocol for Runtime Compliance Evidence and Transaction-Specific Decisioning

cs.CR cs.CY ·2026-10-16 ·Jeremy Blackburn, Tim Hines, Matthew Koepp, Russell Lessard, Jason Patrick
code/data
Abstract · ChainIT white paper CIT-PC-WP-4.0. 32 pages, 5 figures. Companion paper: Provable Authority. Project page: https://chainit.com/white-papers/

Regulated institutions already perform onboarding, screening, monitoring, and review, but compliance evidence ages, and whether it supports a decision depends on the institution, product, subject, counterparty, transaction, jurisdiction, risk tier, and purpose. A relying party therefore often cannot obtain compact, attributable proof of which evidence, policy, authority, and transaction context were current when a decision was made, a gap that grows as authority is delegated to software agents and scheduled workflows. We describe ChainIT Provable Compliance, the compliance-side layer of the ChainIT Authority Protocol. Each sourced or institution-generated fact is represented as a separate Evidence Assertion VDT with provenance, freshness, confidence, permitted purpose, and privacy classification; a versioned Compliance Policy Profile selects the applicable rules; a Business Rules Engine returns deterministic, reason-coded predicate results; and the organization's configured Workflow Engine composes them into a signed, time-bounded Compliance Decision VDT for one reliance event. The decision contributes a single predicate, Compliance Decision Permits, to session-eligibility and transaction-policy decisions. It never independently authorizes payment: only a final Allow together with an exact transaction digest, atomic capacity reservation, and a single-use execution credential can commit value. The paper specifies point-of-reliance freshness, scoped propagation of evidence changes, decision states, evidence domains (AML, sanctions, PEP, background, identity, and licensing), the U.S. and FATF regulatory context as of August 2026, privacy and disclosure controls, and residual risks.

0.45

Recording-Aware Evaluation of Multimodal OT Intrusion Detection

cs.CR cs.LG ·2026-10-16 ·Mustafa Umut Ozbek, Khalil El-Khatib, Li Yang
publishedbenchmarkcode/data
Abstract · 12 pages, 3 figures, 5 tables. Accepted for presentation at the Industrial Control System Security Workshop (ICSS 2026), held in conjunction…

Machine-learning-based intrusion detection systems (IDSs) for operational technology (OT) are commonly evaluated on benchmark datasets in which normal and attack data are collected in separate recordings. In such settings, high performance may reflect recording-associated differences as well as attack behavior. We introduce a recording-aware evaluation procedure that compares attack windows with background from the same recording, uses prior data for model fitting and calibration, tests temporal alignment against randomized event schedules, and reports event coverage together with alarm burden. On the ProvICS multimodal dataset, cyber modalities strongly distinguish the benign and attack recordings before the first labeled attack. Same-recording evaluation reverses the apparent modality ranking across all studied window sizes: the physical view becomes the strongest individual modality, while cyber-only fusion shows the largest difference between cross-recording and same-recording AUC, reaching 0.54. Operational analysis further shows that permissive event-overlap metrics can overstate detection when a small number of long alarm episodes cover multiple actions. An external analysis on HAI shows smaller and reference-dependent recording sensitivity. These findings show that cross-recording benchmark performance should be interpreted alongside same-recording, forward, temporal, and operational evaluations.

0.44

Provable Authority: A Protocol for Verifiable Authority Resolution and Control Across Human, Organizational, and Agentic Transactions

cs.CR cs.MA ·2026-10-16 ·Jeremy Blackburn, Tim Hines, Matthew Koepp, Russell Lessard, Jason Patrick
code/data
Abstract · ChainIT white paper CIT-PA-WP-7.0. 32 pages, 3 figures. Companion paper: Provable Compliance. Project page: https://chainit.com/white-papers…

Payment systems authenticate users, devices, accounts, and cryptographic credentials, but a relying party, such as a counterparty, payment rail, marketplace, or autonomous agent runtime, generally cannot obtain compact, independently verifiable evidence that the party presenting one exact transaction holds current organizational authority to execute it under the applicable evidence, policy, budget, and approval state. Agentic commerce makes this gap operationally urgent: a software agent can produce a valid signature while acting outside its delegated purpose, after its sponsor's authority has changed, or on transaction details that differ from those a human approved. We describe the ChainIT Authority Protocol, a subject-neutral and rail-neutral pre-execution authority and evidence control plane. The protocol separates durable identity, a versioned organizational authority record (the Authority Resolution Pactvera), sourced evidence assertions, deterministic reason-coded policy predicates, and organization-configured dispositions; binds each decision and human approval to a canonical transaction digest; reserves capacity atomically so that limits hold under concurrency; and authorizes commit only through a single-use, sender-constrained Execution Authorization Credential consumed at the rail adapter, with append-only lifecycle evidence sealed after the terminal outcome. An Agent Subject Profile adds the workload registration, attested key hierarchy, and sovereignty boundary required when the acting subject is an AI agent, without introducing a separate authority concept. We specify the commit-time execution predicate and bounded-freshness status rules, give enforcement profiles for fiat (ACH, wire, card, instant payment) and stablecoin rails, and state residual risks and non-goals.

0.42

SPIDES: Information-Flow Control for Browser Agents

cs.CR cs.AI ·2026-10-16 ·Egor Zverev, Juhee Kim, Leon Trampert, Boris Köpf, Aashish Kolluri +2 more
positioncode/data
Abstract

Browser agents automate complex workflows by reading and acting on arbitrary page content, but this flexibility exposes them to prompt injection attacks and unintended data exfiltration. Recent work argues that LLM-driven browser agents erode traditional web security boundaries such as the Same-Origin Policy (SOP) by acting across origins with the user's authority and no built-in isolation. We argue that restoring SOP is neither sufficient (a same-origin public sink can leak private data) nor necessary (legitimate tasks require authorized cross-origin flows) for protecting the security of users. We propose SPIDES, which instead enforces information-flow control by gating primitive browser actions such as typing, clicking and navigating on labels propagated from element-level HTML confidentiality and integrity annotations. Untrusted and confidential content is masked from the agent's context, while policies block unauthorized flows, providing security guarantees against both prompt injection and data exfiltration that hold independently of the model driving the agent. To evaluate SPIDES and baseline defenses under a common threat model, we build MicroNet, a closed ecosystem of 28 interlinked websites with element-level ground-truth labels and paired benign/adversarial tasks. SPIDES achieves 0% attack success rate by design while retaining high task completion rate and autonomy. Our findings demonstrate that information-flow control is an effective technique to extend web security to browser agents, gaining on autonomy and utility against state-of-the-art baselines. The source code and data are openly accessible at https://github.com/microsoft/SPIDES.

0.41

When Core-Network Elements Become Reachable: Field-Level Exploitability Analysis in Diameter

cs.CR ·2026-10-16 ·Wei Yi, Changlong Han, Xue Zhang, Qiao Zhang, Yi Chen +5 more
code/data
Abstract

The increasing exposure of cellular core-network elements to public networks has turned Diameter signaling interfaces into a practical attack surface. While large language models (LLMs) show promise for protocol exploitability analysis, we find that their judgments on Diameter can be unreliable because they may deviate from the tacit analysis discipline followed by security experts and reach conclusions without considering complete protocol evidence. To address this challenge, we present LENS, an LLM-based system for reliable field-level Diameter exploitability analysis. LENS guides LLMs using both exploitability-analysis discipline and the protocol evidence needed for analysis. It infers analysis discipline from existing exploit writeups and evidence-collection guidance from exploitability-analysis failures, enabling systematic evidence collection, exploitability assessment, and proof-of-concept (PoC) attack generation when applicable. Our experiments show that LENS performs reliable analysis across four LLMs and repeated runs, achieving mean recall of 94.00% to 98.00%, precision of 78.73% to 88.94%, and accuracy of 84.00% to 92.00%. We find that more than 25% of field paths in the S6a and Cx interfaces are exploitable under the considered threat model. We further assess their impact on practical implementations and uncover 95 previously unknown vulnerabilities, including 17 in open-source EPC/IMS implementations and 78 in commercial deployments, causing service disruption, information leakage, subscriber tracking, and accelerated battery depletion. Applying LENS to 16 additional protocols spanning cellular, network, industrial, and host-internal systems uncovers 118 previously unknown vulnerabilities, demonstrating its strong generalizability.

0.38

TUPLE-Bench: Evaluating and Mitigating Privacy-Utility Trade-offs in Typed Decision Models

cs.CR cs.AI ·2026-10-16 ·Yiqu Yang, Qiaoyuan Zheng, Yang Tian, Yu Fan, Elliott Ash
benchmarkhot direction
Abstract · Our public website: https://qiaoyuan-zheng.com/TUPLE-Bench/

Typed decision models often return more than a selected category. They also expose probabilities for each candidate and, sometimes, a separate confidence score. These fields support downstream decisions but may reveal information beyond a recipient's authorization. We introduce TUPLE-Bench (Typed decision models' Utility and Privacy Leakage Evaluation Benchmark), a benchmark of paired synthetic cases in which recipients may learn a coarse operational category but not its fine-grained cause. Across nine models, instructions to withhold the cause fail to protect it: 91.88%-99.93% of responses reveal the cause through direct selection or residual scores. Even when causes are excluded from the candidate set, full outputs enable 62.85%-75.35% recovery on held-out cases, compared with 51.04%-57.64% using only the selected category (50% chance). Full outputs also reduce mean downstream cost in all 27 model-task comparisons. However, confidence fields increase leakage without consistently reducing cost. We further introduce workflow-constrained privacy quantization (WCPQ), which reduces secret recovery while preserving the behavior and cost of specified downstream workflows. Our findings highlight the need to audit decision interfaces field by field and workflow by workflow.

0.35

ToolRosetta: Surface-Form Invariance of Tool-Call Guards

cs.CR cs.LG ·2026-10-16 ·Hasan Demirkiran, Jens Ernstberger
publishedbenchmark
Abstract · Accepted at the NeurIPS 2026 Workshop on Agents in the Wild: Safety, Security, and Beyond (AIWILD). 4 pages plus appendices

Step-level guards for tool-using agents are evaluated on tidy JSON tool calls, but deployed coding agents act through execute_bash and execute_code tools whose payloads are strings. We introduce ToolRosetta, which re-renders every candidate action in the AgentDojo and AgentHarm splits of TS-Bench as a Python call, as a shell command, and without the agent's rationale. The renderings are produced by code and verified by round-trip parsing, so the user request, the history, the label and the action's effect stay fixed while only the surface form changes. Across eleven guards with two placebo controls, frontier and strong open-weight guards are stable: no frontier error rate moves by more than 3.4 points. Four of five specialized agent guards are not: under the shell form their false-positive rates rise by 21 to 47 points on AgentDojo, and TS-Guard's by 38 points on its own home benchmark, blocking between a third and nine in ten benign actions; the fifth, StepGuard, does not move. A small open guard drifts five points more permissive. Removing the agent's rationale moves most guards whose input carries it, in guard-specific directions. A request-blind control on one guard shows that the rationale is where intent is read from. A parser that restores the canonical call in front of the guard removes the effect for these renderings, by construction. Benchmark accuracy on JSON calls does not predict a guard's operating point behind a code-acting agent.

0.34

HardEval: An Execution-Verified Benchmark for AI-Driven Security Hardening and Compliance Remediation

cs.CR cs.LG ·2026-10-16 ·Manohar Vellala, Soumya Banerjee, Sandip Roy, Ross Gore, Sachin Shetty
benchmarkcode/data
Abstract · 8 pages, 4 figures, 4 tables. Code and data: https://github.com/manoharvellala/Hardeval The paper has 4 figure environments (two are images,…

Implementing Security Technical Implementation Guide (STIG) controls remains a demanding operating-system administration task. Each natural-language security requirement must be translated into a precise and persistent configuration change. These changes span Secure Shell (SSH) configuration, authentication policy, audit rules, and kernel parameters. Existing benchmarks do not measure whether a model can perform this task on a live system. Code benchmarks evaluate isolated programs without persistent system state, while agent benchmarks largely focus on applications rather than operating-system configuration. In this paper, we introduce \textbf{HardEval}, an execution-verified benchmark for evaluating language-model-assisted security hardening. HardEval is built from Red Hat Enterprise Linux 8 STIG controls that fail on an unhardened baseline system. Expert-written reference remediations and Open Vulnerability and Assessment Language (OVAL) checks serve as hidden ground truth for grading; they are never shown to the model, and each task states only the required security outcome while withholding the implementation. The model produces a Bash script, which is executed on a live virtual machine and graded from fail to pass using the corresponding OpenSCAP/OVAL compliance check. Evaluation of fourteen frontier and open-weight configurations reveals large and uneven capability differences across security domains. A recurring failure mode is plausible security code that does not satisfy the exact compliance condition. For four open-weight configurations, multi-run Wilson confidence intervals preserve the observed ordering. HardEval is publicly available.

0.32

Task-aware Kernel Stein certificates for Rényi differential privacy

cs.CR stat.MLcs.LG ·2026-10-16 ·Leoni Carla Wirth, Tara Trauthwein, Gesine Reinert
Abstract

Rènyi differential privacy, based on Rènyi divergence, has become a notion of choice for privacy guarantees. However, outside the Gaussian setting, Rènyi divergence can be difficult to compute. In particular, Rènyi differential privacy guarantees that exploit task-specific information, as in Rènyi Pufferfish privacy, can be challenging. Here we use kernel Stein discrepancies based on escort distributions to provide certificates for Rènyi differential privacy. By designing the kernel to focus on the relevant privacy directions, we also derive task-aware privacy certificates. We provide exact identities for Gaussian and Laplace mechanisms and bounds for exponential mechanisms, with separate task-aware guarantees. A real data set example illustrates how the protected comparison set and the released subspace jointly determine the resulting certificates.

0.32

SAFE-Mem: Secure Agentic Memory with Fine-Grained Authority and Governance-Aware Retrieval

cs.CR cs.CL ·2026-10-16 ·Haoming Li, Dingyi Kang, Xu Hu, Jinxin Yang, Zhu Xu +6 more
code/data
Abstract

Large language model (LLM) agents increasingly rely on persistent memory to support long-term reasoning and personalized interactions. However, information collected from external sources can introduce security risks when untrusted claims (i.e., atomic statements) are later treated as authorized preferences, policies, or permissions. Existing defenses typically rely on content-based filtering or coarse-grained provenance tracking, which either fail to prevent unauthorized memory use or block legitimate information, sacrificing memory utility. To address these limitations, we propose SAFE-Mem, a fine-grained memory governance framework that improves security while preserving legitimate memory use. First, SAFE-Mem annotates each claim with three governance labels (i.e., kind, authority, and scope) to capture its provenance and permitted use. Second, governance-aware retrieval matches query requirements with claim permissions to block unauthorized influence while retaining legitimate information. Third, a governance-enabled memory graph connects source sentences, claims, and entities to support provenance tracking and efficient retrieval. Across four LLM backbones on MPBench and LoCoMo, SAFE-Mem attains the lowest attack success among defenses that retain legitimate memory, at least 77% below the strongest of them, and the highest answer accuracy of all evaluated defenses, improving it by up to 15 points. The code is available at https://github.com/LVenum/SAFE-Mem.

0.32

Token-Level Attribution of Multi-Mode LLM Watermarks: Optimal Rule and Belief Weighting

cs.CR eess.SPstat.AP ·2026-10-16 ·Rui Zhou, Licheng Zhao, Chao Shen, Wenqiang Pu
Abstract · 13 pages, 9 figures

A watermark verifier may hold many modes, i.e., secret keys, and a document may mix spans of different modes with unwatermarked text. Directly reporting the mode with the largest pivotal statistic (pivot) inflates the Type I error, and restoring the nominal level reduces power as candidate modes are added. We formulate token-level multi-mode attribution for the Gumbel-max watermark as maximizing correct-label power under a Type I error constraint given a mode prior. We show that the optimal rule reports the mode with the largest prior-weighted alternative density and rejects when this density exceeds one common threshold. Based on this rule, we propose the Belief-weighted Likelihood Attribution with Density Estimation (BLADE) detector. It estimates the mode prior from neighboring pivots and the alternative density from reference token probabilities. Under a conditional product model of the pivots, it keeps the exact Type I error. We further show that BLADE approaches the optimal rule as its prior and density estimates improve, with a sharp coefficient on the density error. On OPT-1.3B and other open-weight large language models, BLADE attains higher correct-label power than the compared detectors across candidate counts, token-substitution rates, generation temperatures and generator models.

0.31

GRAD: Graph-Structured Antibody Defense for LLM Agents

cs.CR ·2026-10-16 ·Tianqi Song, Jiawei Du, Xuemei Jia, Zheng Wang
Abstract · 30 pages, 2 figures, including appendices

Memory poisoning and decomposition attacks distribute malicious evidence across the operations of large language model agents. Existing request-level defenses can miss such attacks because they assess individual inputs without connecting earlier information to later actions. Moreover, detectors must adapt as task contexts change, while updates to shared parameters can impair previously learned detection capabilities. To address these challenges, we propose GRAD, which combines graph-based evidence aggregation across operations with an extensible collection of detection experts. Its Semantic-Structural Understanding Module (SSUM) jointly encodes operation text, event attributes, and typed candidate relations in an execution graph to aggregate distributed attack evidence into a graph-level risk score. Its Continual Antibody Adaptation (CAA) adds experts on frozen encoder snapshots and routes inputs by observable graph signatures, acquiring and reusing detection capabilities as tasks arrive. Across Decomp, MCP, and MRT, GRAD achieves AUROCs of 97.19%, 77.59%, and 86.60%, respectively, outperforming Sequential Monitor and locally adapted Agent-Sentry and NeuroTaint. CAA improves final accuracy by 13.08 percentage points over Replay on the cross-dataset task stream. The gain remains 9.05 points with equal family/base-case weights within each task. Its compact drop-in gate performs local model inference in milliseconds without monitoring API calls.

0.30

Aspire: Practical Batch Queries on Dynamic Encrypted Database with Full Leakage Suppression

cs.CR ·2026-10-16 ·Hao Nie, Wei Wang, Peng Xu, Haoming Yan, Laurence T. Yang +1 more
hot direction
Abstract

Leakage-abuse attacks against encrypted database systems exploit semantic leakage and pose a severe threat to the confidentiality of plaintext data and queries. To address this problem, we propose a zero-leakage encrypted database that rules out leakage-abuse attacks, which fundamentally rely on semantic leakage. We first distinguish between two kinds of zero-leakage security. Ideal zero-leakage forbids the disclosure of any information beyond public parameters, whereas restricted zero-leakage additionally permits the leakage of metadata (such as database size and query timing). We show that restricted zero-leakage is sufficient to thwart leakage-abuse attacks and is achievable in practice. We then formally define adaptive restricted zero-leakage security and present Aspire, an encrypted database that satisfies this notion. Aspire integrates probabilistic batch code and minimal perfect hashing to support efficient batch keyword queries and dynamic updates while revealing only non-semantic metadata. Experimental results demonstrate the practicality and efficiency of Aspire, showing that Aspire can achieve an amortized query latency of less than 1.38 ms (7.53 s for a batch of 5461 keyword queries), with an amortized communication cost of approximately 2.95 KB (16.1 MB in total), on a $2^{22}$ key-value pairs database.

0.30

ABF-Bench: A Verifiable Benchmark for Agentic Blockchain Forensics

cs.CR ·2026-10-16 ·Xiaohang Yu, Hejia Geng, William Knottenbelt
benchmark
Abstract

Blockchain forensics is adversarial and open-ended: money launderers split, swap, and bridge funds, cut their trail at mixers or custodial services, and change tactics from one incident to the next. Investigators must keep adapting their tracing in response. Yet the immutability of public ledgers preserves every recorded movement in public view, making much of the investigation independently verifiable. We introduce ABF-Bench, an interactive benchmark that makes evidence-backed fund tracing a verifiable agent task, in which every claimed movement must cite ledger records the agent retrieved. A case-to-task pipeline builds well-posed tasks across ledgers: conservative rules halt tracing wherever the records leave a continuation unresolved, and each task's reference graph is the answer those rules determine. Built from real laundering activity, the tasks combine multi-hop transfers, DeFi exchanges, cross-chain round trips, custodial and mixer deposits, and address poisoning. We also propose the ABF agent harness, built around a working memory, and evaluate it alongside general-purpose agents and fixed-order tracers. The harness reaches 75% edge recall, against 47% for Codex with the same model and 10% for Claude Code, which loses recall to overconfidence: it cites records it left unretrieved. Compared with fixed-order tracers, the harness acquires evidence faster than the best of them early in the budget and recognizes undisclosed custodial services that they walk past (89% vs. 0%), with 98% fidelity. However, agents tend to stop early and lose recall on larger tasks, which holds their final recall below the best tracer. Beyond exposing these gaps, ABF-Bench opens blockchain forensics to LLM agents as a verifiable domain, where the verdicts that score agents also offer potential training signals.

0.29

RATION: Towards Quantitative Risk-Based Access Control for AI Agents

cs.CR cs.AIcs.LG ·2026-10-16 ·Yuntao Du, Harsh Chaudhari, Yuetian Chen, Ron Heichman, Ashish Jayamohan +6 more
code/data
Abstract · Code is available at https://github.com/perplexityai/RATION

AI agents increasingly interact with external environments to complete complex user tasks. These interactions can lead to unexpected behavior and expose agents to security threats such as indirect prompt injection. Current agent security systems largely protect agents by restricting their actions, which can result in a substantial security-utility trade-off for complex, open-ended tasks. To address this, we propose a quantitative risk-based approach to access control for AI agents that regulates measurable risks accumulated during task execution. As a concrete realization of this approach, we introduce RATION, a security framework that tracks measurable risks associated with agent actions and enforces quantitative constraints throughout execution. RATION first constructs a task-specific policy by combining explicit constraints derived from the user request with implicit risks calibrated from historical executions of similar tasks. During task execution, RATION continuously tracks the accumulated risks and applies different enforcement strategies depending on the type of risk. Extensive experiments on four security benchmarks show that RATION provides strong security while maintaining high task utility across diverse models and environments.

0.29

Unbounded Losses, Unchecked Inputs: Measuring the Safeguards of LLM Options-Trading Agents

cs.CR cs.AI ·2026-10-16 ·Mengxiao Wang, Nitesh Saxena
code/data
Abstract · 26 pages, 9 figures, 8 tables. Original measurement study: a logged screen of 765 GitHub repositories and cell-level scoring of 26 LLM optio…

Large language model (LLM) agents now place options orders at retail brokers, and developers publish open-source options agents faster than anyone has examined them. Prior security work on trading agents scored academic equities designs by whether an attacker can flip a buy-or-sell decision. We present the first measurement of the options agents that brokers and developers ship, scored by the safeguards they enforce and by the worst loss their documented order paths can reach. We screened 765 GitHub repositories, broker documentation, and the research literature and found 215 eligible repositories and 8 broker-native agent interfaces. We then scored 26 systems, including every broker interface and every open-source project with at least 20 stars, on seven robustness and eleven security safeguards with cell-level evidence. The ecosystem is new: 198 of the 215 repositories were created in 2026, and 129 (60.0%) appeared in the two weeks around one options-trading hackathon. Of the 22 systems that can place orders, 10 enforce a pre-trade limit outside the model, yet none of the 26 authenticate their text sources or cross-check market data. Of the 26, 17 can execute with no mandatory human step and 13 can reach unbounded loss, including 6 of the 8 broker interfaces. Because an attacker who writes the agent's inputs can move an order's strike, expiry, volatility, and legs while its direction stays correct, options agents need loss limits enforced at the order boundary.

0.28

Empirical Analysis of TEE-Enabled Rollup Designs

cs.CR cs.DC ·2026-10-16 ·Claudiu-Nicu Bărbieru, Corentin Libert, Peterson Yuhala, Andrei Arusoaie, Matthieu Bettinger +5 more
published
Abstract · Accepted for the 24th IEEE International Symposium on Network Computing and Applications, NCA 2026; Copyright is with IEEE

Layer 2 (L2) rollups have become the dominant scaling solution in Ethereum. At the core of these systems lies the rollup sequencer, a component that orders, executes, and submits L2 transaction blocks to the main Ethereum blockchain. The centralization of this sequencer introduces security risks, including censorship and malicious manipulation of transactions to extract maximum extractable value (MEV). Trusted execution environments (TEEs) provide strong hardware-enforced confidentiality and integrity guarantees for code and data. TEEs are strong candidates for strengthening security and ensuring verifiable neutrality in rollup sequencer operations, but multiple TEE-based designs are possible, offering different trade-offs among modularity, performance, and security guarantees. Among them, the separation of proposer and builder logic (PBS) into independent services is a strong contender against the monolithic approach that bundles all sequencer logic. Despite several proposed theoretical frameworks, there is a lack of experimental evaluation of the practicality and security trade-offs of integrating TEEs into rollup sequencers. This work addresses this gap through a comprehensive empirical study of TEE-enabled rollup designs under both PBS and monolithic sequencer models. Our experimental evaluations show that the impact of TEEs on rollup performance is small and that PBS does not significantly amplify it. We also show that PBS can lead to significant performance improvements by enabling more specialized, efficient builder logic implementations.

0.27

Stealthy Poisoning Attacks via Reused KV Cache of Reasoning

cs.CR ·2026-10-16 ·L. Zhu, Y. Jung, S. Szyller
Abstract · 18pages, 2figures

Modern large language model (LLM) systems reuse key-value (KV) caches to accelerate inference. Recent work has shown that reusing KV caches from reasoning traces can also improve task performance, similar to retrieval-augmented generation (RAG). We demonstrate that this form of KV cache reuse introduces new poisoning vulnerabilities. We propose R$^2$ECAP (Reused Reasoning-Cache Poisoning), a black-box attack that plants poisoned reasoning traces solely through interactions with a system's chat interface. R$^2$ECAP constructs two traces: a preceding trace that steers the model toward an adversarial target and a retained trace that carries this target. Because the retained trace is generated conditioned on the preceding trace, its KV cache preserves residual adversarial influence, even after the preceding trace and its cache entries are discarded. We evaluate R$^2$ECAP on three open-weight LLMs and three datasets. R$^2$ECAP achieves up to 79% attack success rate (ASR) and outperforms HijackKV, a state-of-the-art white-box KV poisoning attack, in 8 of 9 resource-constrained settings. It also achieves 46% ASR in a realistic reasoning-trace retrieval system. Finally, R$^2$ECAP remains stealthy, with only 1% of its retained responses detected by a popular prompt injection detector.

0.26

Automated Rule Mutation for Emerging Attack Variants using Infrastructure-Aware Technique Proximity

cs.CR ·2026-10-16 ·Sepehr Ghaffarzadegan, Boubakr Nour, Makan Pourzandi, Mourad Debbabi, Chadi Assi
Abstract · Submitted for publication and currently under review

Detection rules are typically crafted based on specific observations of an attack. While these rules are effective for particular attack instances, their coverage diminishes when adversaries modify their techniques, such as by employing alternative execution paths or adapting to different telemetry environments while achieving the same objective. To address this challenge, we introduce ARMS, an automated rule mutation solution that generates variant detection rules. ARMS defines a novel three-dimensional proximity space to identify techniques related to the original attack that capture structural relations, functional similarity, and environmental constraints. Rather than relying on ungrounded generation, ARMS leverages verifiable attack execution commands, which are extracted via semantic alignment with the Atomic Red Team repository, to construct a context-aware Retrieval-Augmented Generation (RAG) pipeline. We evaluated ARMS using verified, public SigmaHQ rules and successfully generated variants, demonstrating ARMS's capability to expand detection coverage. We further evaluated ARMS across multiple research questions, showing the effectiveness of each proximity dimension and infrastructure-aware rule customization. Nearly all generated variant rules successfully compiled into SIEM queries, and ~92% of the newly generated rules detect the original behavior despite attackers' technique variations.

0.26

On the (In)stability of Adaptive Fee Mechanisms in Multi-Chain Systems

cs.CR cs.DC ·2026-10-16 ·Yifan Zhou, Jiang Xiao, Xuechao Wang, Ru Cheng, Shijie Zhang
unusual
Abstract

EIP-1559 has become the dominant transaction fee mechanism in permissionless blockchains, regulating scarce block space through an adaptive base fee. The design balances a trade-off between responsiveness to demand fluctuations and stability of long-term resource usage. On a single chain, this trade-off is well understood and empirically stable. In multi-chain systems, however, cross-chain flows introduce delayed economic interactions across otherwise independent fee markets. These interactions alter the control dynamics and create stability risks that existing models do not capture. Deployed systems have already polarized. Polkadot enforces conservative update rates, effectively disabling fee adaptivity and reverting the market to inefficient first-price auctions. Cosmos adopts aggressive update rates without formal guarantees. To fill this gap, we propose a control-theoretic framework that formalizes multi-chain fee markets as a nonlinear dynamical system with delayed feedback. We identify delay-amplified over-correction as the structural root of instability. Experiments show that cross-chain coupling erodes stability relative to monolithic blockchains, reducing the safe parameter space for base fee update rates by up to 84.8%. Moreover, instability propagates across chains, contaminating the otherwise stable peer chains. At Ethereum's update rate, which leaves a monolithic chain stable, the multi-chain system exhibits extreme load oscillation between full and near-empty blocks, with over half of all blocks at full capacity. Mainstream Cosmos chains already exceed their safe bounds under their default configurations, and the resulting load oscillation raises their block failure rate by an order of magnitude. Guided by this mechanism, we design a lightweight protocol-level mitigation that removes the over-correction and stabilizes dynamics in multi-chain environments.

0.25

Surprise Masking: Private Data Filtering with Model Likelihoods

cs.CR ·2026-10-16 ·Riya Tyagi, Sahra Ghalebikesabi, Vinnie Monaco, Avi Schwarzschild, Om Thakkar
Abstract

Data useful for language model (LM) training can contain private or unwanted information that is at risk of being learned or memorized. We propose Surprise Masking (SM), a simple form of perplexity filtering that uses a publicly available reference model to identify unwanted spans and suppress learning from them while preserving the rest of each example. On synthetic code benchmarks, we show SM improves the privacy-utility tradeoff over standard masking methods and PII filtering. In fine-tuning experiments, SM reduces approximate extraction of private values from revealed prefixes from 66.3% to 2.3% and the reproduction of stylistic coding quirks from 23.6% to 1.3%, both while preserving performance on the evaluated coding tasks. We also evaluate SM for sanitizing datasets by directly removing unwanted content. Here, SM outperforms DePA, a perplexity-based defense for detecting poisoned dead code, while specialized detectors perform better on broader credential and PII benchmarks. Across eleven settings, we find that SM's downstream success is predictable from the separation of unwanted and useful content in the distribution, and it is particularly useful with code data. Ultimately, SM is an inexpensive, adaptable method to preserve valuable examples while reducing the learning of unwanted information.

0.20

Veil-Then-Reveal: Protecting PII Privacy via Reversible Pseudo-Anonymization in Zero-Trust User--LLM Interaction

cs.CR cs.CL ·2026-10-16 ·Mohamed Elmahallawy, Mohamed Shaaban, Thomas E. Carroll
Abstract

Large language models (LLMs) increasingly process sensitive documents and personal information, yet sending prompts containing personally identifiable information (PII) to remote providers exposes data subjects and organizations to risks from logging, retention, model training, and unintended disclosure. Existing anonymization methods often mask or remove PII, while conventional encryption eliminates the lexical and structural cues needed for effective LLM processing. We present Veil-then-Reveal (VtR), a user-side PII pseudo-anonymization framework for privacy-preserving LLM interaction under a zero-trust threat model. VtR treats remote LLM providers as untrusted and sends only pseudo-anonymized prompts, keeping plaintext PII on the user. VtR first detects PII and replaces it with realistic, same-type, in-vocabulary surrogates before prompts leave the user's device, preserving their linguistic and structural properties. It combines format-preserving encryption (FPE) for structured PII, such as phone and credit-card numbers, with a keyed-hash dictionary and secure vault for semantic PII, such as names, locations, and organizations. Session-specific mappings provide cross-session unlinkability, while user-side restoration reversibly replaces surrogates with their original values in LLM responses. We further introduce a rank-class FPE strategy for dates that preserves format and ambiguity while reducing cipher evaluations. Evaluation on three datasets containing 11,322 annotated PII instances achieves 97.9% detection recall, 95.2% protection of detected values, and 99.9% round trip; 100% of echoed surrogates end-to-end round-trip restoration.

0.20

Under the Skin: Backdoor Infection in Multimodal Large Language Models

cs.CR ·2026-10-16 ·Jiawen Wang, Ruo Yang, Haoran Dai, Binghui Wang, Ning Luo
Abstract

Multimodal Large Language Models (MLLMs) have achieved impressive performance in generating text conditioned on inputs from different modalities. In this work, we introduce backdoor infection, a cross-modal attack framework targeting MLLMs. Starting from a model already injected with a backdoor in one modality, our attack constructs a corresponding trigger in another modality, called the infected modality. Exploiting the model's shared latent space, the trigger in the infected modality reliably activates the malicious behavior, without modifying or poisoning the infected modality during training. We further develop a lightweight, constraint-aware optimization method for efficient and practical trigger generation. Extensive experiments across several state-of-the-art MLLMs, including LLaVA, InternVL, Qwen-VL, and Gemma, demonstrate that our approach identifies cross-modal triggers that are more efficient, effective, and less detectable than direct optimization baselines, highlighting a critical new threat vector in multimodal systems.

0.19

From Billion-Scale to Million-Scale Language Models: Operational Capabilities for Embedded Post-Exploitation

cs.CR cs.AI ·2026-10-16 ·Gabriel Hortea, Simon Pietro Romano, Juan Tapiador, Mario Varlese
code/data
Abstract

Small Language Models (SLMs) are increasingly attractive for resource-constrained environments due to their low computational footprint. These characteristics also make them candidates for future autonomous malware, where connectivity, latency, privacy, or operational constraints may preclude cloud-based inference. Yet the ability of ultra-compact models to support malware during post-exploitation remains poorly understood. We systematically evaluate seven open-source SLMs (20M-1.5B parameters) on two post-exploitation code-generation tasks: Directory Enumeration and Backdoor Generation. We compare Python (high-resource) and Lua (low-resource) across prompting strategies, task formulations, iterative reasoning, grammar-constrained decoding, and execution-feedback loops. Results show that offensive code-generation capabilities do not increase linearly with model size. Successful execution depends on the interaction of model capacity, task formulation, and prompting strategy, with task formulation exerting a greater influence than prompt engineering alone. Algorithmic formulation combined with few-shot prompting enables Qwen2.5-Coder 1.5B to reach Pass@5 of 100% in both Python and Lua for Directory Enumeration, 100% in Python for Backdoor Generation, and 99.2% in Lua. Although the evaluated models remain too computationally demanding for widespread deployment in contemporary malware, rapid progress in compact models may reduce this barrier. These findings help anticipate AI-assisted malware and inform future countermeasures.

0.19

MobileJailBench: Red Queen Jailbreak Benchmarking for Mobile Agents

cs.CR ·2026-10-16 ·Songping Wang, Borui Li, Qinglong Liu, Yueming Lyu, Yi Huang +3 more
benchmark
Abstract

MLLM-based mobile agents are rapidly gaining autonomy in perception, planning, and closed-loop execution in dynamic mobile environments. However, this autonomy also broadens the attack surface and amplifies the risks of malicious misuse, including privacy leakage, device compromise, and large-scale phishing. Existing jailbreak evaluations leave three critical gaps for mobile agents: they focus on conversational attacks instead of mobile workflows, use static task sets with fixed difficulty, and rely narrowly on attack success rate rather than comprehensive security assessment. To close these gaps, we present MobileJailBench, the first evolving jailbreak benchmark for mobile agents, comprising three key components: (1) a benchmark comprising 520 tasks spanning three jailbreak paradigms and 14 risk categories, built on a high-fidelity mobile sandbox that we construct, enabling realistic workflows across over 20 widely used apps; (2) Dual-Dynamic Evolution, which combines Failure-Distilled Jailbreak Evolution with Weakness-Driven Benchmark Expansion; and (3) Stratified Adjudication Framework for assessing security awareness, execution behavior, and risk consequences. Extensive experiments show that current mobile agents are particularly vulnerable to implicit jailbreaks, as their safety alignment remains largely superficial and fails to generalize from conversational refusal to dynamic mobile interactions. Moreover, our dynamic evolution mechanism sharpens practical threat boundaries by continually uncovering new vulnerabilities, reaching a 100% ASR on Qwen3.5-122B. MobileJailBench provides a reproducible, continuously evolving framework for tracking the true security boundaries of mobile agents.

0.17

MobileInject: Evaluating Reasoning-to-Action Security of MLLM-Based Mobile Agents under Cross-Modal Consistency Injection

cs.CR cs.AI ·2026-10-16 ·Songping Wang, Peiyang Li, Yi Huang, Fengting Li, Xuhao Ran +3 more
benchmark
Abstract

Multimodal large language model (MLLM)-based mobile agents are rapidly evolving from conversational assistants into autonomous systems that perceive, reason over, and act on smartphone environments. This capability also exposes a new security risk: adversaries can embed malicious objectives in cross-application mobile content and steer agent behavior through indirect multimodal injection. Although prompt injection has been studied in desktop and web settings, its cross-modal properties in mobile ecosystems remain poorly understood, and existing security analyses are still limited. To close this gap, we introduce MobileInject, the first benchmark for evaluating mobile agents against multimodal injection attacks and reasoning-to-action security risks. MobileInject provides a high-fidelity interactive mobile evaluation framework covering 11 scenarios, 9 security-risk categories, and 108 executable risk instances. It further introduces Cross-Modal Consistency Injection to exploit corroboration across modalities, two enhanced attack methods, Multi-Role Collaborative Deception and Reasoning-Camouflage Injection, to intensify trust manipulation and reasoning hijacking, and a Reasoning Security Diagnosis Framework for fine-grained failure analysis. Systematic experiments show that mobile agents are broadly vulnerable to multimodal injection, and that many failures stem from reasoning hijacking at early stages. These findings establish MobileInject as a systematic cross-modal benchmark for studying and improving the security of mobile agents. Code release details are provided in the Appendix Open Science section.

0.17

AgentLoupe: Kernel-Resident Semantic Causal Labels for AI Agent Observability

cs.CR ·2026-10-16 ·Yifei Bao, Ting Liu, Yuxin Ren, Ruizhe Huang, Likang Luo +3 more
Abstract · Accepted to the 5th Workshop on Practical Adoption Challenges of ML for Systems (PACMI 2026), co-located with ACM SOSP 2026

AI agents translate user intents into system actions via tool invocations. Accurately attributing these actions to their originating LLM interactions is essential for security auditing, yet existing methods, such as AgentSight, fail for delayed execution and cross-process-lineage behaviors. We present AgentLoupe, an application-transparent eBPF-based observability tool. The core insight of AgentLoupe is to embed causal identity into kernel resources. AgentLoupe employs eBPF to capture TLS plaintext to interpret LLM interactions, and assigns each a kernel-resident semantic label. It maintains labels in eBPF maps, then propagates this label across processes and threads, associates it with kernel resources such as files and sockets. When a persistent resource carries context to an independent process outside the original lineage, this design preserves the delayed causal association. Experiments show that the average observation overhead of AgentLoupe is below 8%, and case studies confirm AgentLoupe traces cross-lineage persistent attacks back to their originating LLM interactions.

0.16

When Approval Becomes Execution: Authorization-Grounded Workflow Attacks on Tool-Using Agents

cs.CR ·2026-10-16 ·Chenggong Shi, Hongmei Mou
benchmark
Abstract · 7 pages, 1 figure, 7 tables

Tool-using language-model agents increasingly rely on local services for document access, approval management, and enterprise workflow execution. This paper studies an authorization-grounded workflow attack in the OpenClaw Agent Security Attack Competition, a controlled benchmark derived from SeClaw-style security task synthesis. Rather than using explicit policy-override instructions, the attack guides the agent through a locally coherent approval workflow: a protected document request is denied, the agent requests a normal approval token from the local approval service, and the returned token is supplied to the document service through authorized_by. Official submission traces show this pattern succeeding on three Enterprise OA document targets. We also conducted 23 controlled local validation experiments in the official OpenClaw execution environment: the original configuration reached the approval-token transition in 9/9 reruns and completed the exact target operation in 8/9 reruns, while single-component variants remained successful on the three document targets. The results provide concrete case evidence of an authorization-context gap: token possession can be treated as sufficient authorization even when requester identity, approver authority, resource scope, action type, destination or recipient, and business purpose are not jointly enforced. We present the attack pattern, controlled trace analysis, and defensive implications for context-bound tokens, service-side policy binding, agent-side authorization checks, and trace-aware benchmark design.

0.14

Trojan Horse: Automated Repository Poisoning against Coding Agents

cs.CR ·2026-10-16 ·Xiao Yang, Jinchang Hou, Jiawei Chen, Zhonghou Lv, Junyuan Shang +1 more
benchmark
Abstract

Coding agents are increasingly capable with autonomy, which exposes them to repository-level poisoning. Existing red-teaming studies remain restricted to manual selection for malicious payload locations and fixed disguise strategies, which cannot represent realistic repository poisoning. We present TrojanForge, the first automated framework for repository poisoning. TrojanForge parses repository structure through abstract syntax trees (AST) and maps agent execution traces onto AST nodes. A security review agent ranks these candidates and selects the least conspicuous ones for payload mounting. Then, through adversarial play with the target agent, TrojanForge iteratively refines disguise strategies by inverting the causes of agent detection, and transfers accumulated success experience across repositories. We further construct TrojanBench, a benchmark of 384 poisoned repositories spanning 6 domains and 9 programming languages. Extensive experiments show that TrojanForge achieves a stable poisoning success rate of over 60% with backdoor injection exceeding 90%. Comprehensive evaluation on TrojanBench across 4 harnesses and 12 models reveals that even commercial agents fail to defend against repository-level poisoning.

0.11

RAMP: Toward Ubiquitous Quantum-Safe Key Distribution as a Service

cs.CR ·2026-10-16 ·Jiafan Wang, Sebastian Kish, Dongxi Liu
Abstract

Quantum key distribution (QKD) offers a security foundation independent of computational hardness assumptions. However, its reliance on specialized hardware limits access by ordinary users. Turning QKD into a service poses a fundamental trust challenge, as the providers that supply QKD-derived material themselves could retain knowledge of the keying inputs of users. This makes end-to-end confidentiality solely depend on the trustworthiness of the service infrastructure. We present RAMP, a service-oriented framework that decouples QKD material generation from key consumption. Providers operate QKD infrastructure and supply raw-key materials, while users securely acquire, mask, and locally process them into final keys without operating QKD devices. RAMP employs reconciliation-preserving masking compatible with both DV- and CV-QKD. It mitigates the trust dependence by separating the QKD-service and user-side masking domains. We formalize complementary confidentiality guarantees: RAMP protects the final key even if colluding providers expose their complete raw-key views, provided the masking domain remains secure; conversely, exposing the complete masking state does not break final-key secrecy as long as the QKD source and acquisition paths remain protected. The results establish the dual-domain security boundary of RAMP. We implement RAMP using experimentally acquired QKD data and standardized quantum-safe cryptography. After acquiring service-provisioned keying material, commodity endpoints complete user-side QKD post-processing with an average key-distillation throughput of 2.1Kb/s on an M4 Pro CPU. To our knowledge, we demonstrate the first experimental realization of such service-provisioned, user-side QKD processing, presenting a viable path toward QKD as a service.

0.11

Red-Teaming Multi-Agent Attacks for Loss-of-Control Risks

cs.CR cs.AIcs.MA ·2026-10-16 ·John Kitaoka, Thao Amelia Pham, Carlos Rafael Giudice, Jacob Merizian
Abstract

AI agents are growing more capable and persistent, and are being deployed in ever larger numbers. As more start collaborating, AI control measures must remain effective when they covertly collude on a harmful objective. We introduce MiniLab, a synthetic series of live services and codebases (5 interconnected services, 249 benign tasks, 8 attack objectives), adapted from internal AI deployment infrastructure. We demonstrate distributed attacks, in which different agents carry out separate attack fragments, and how their covertness evades frontier monitors when reading their trajectories individually. We show how distributed attacks spread across diverse contexts and harmful objectives through 16 human-optimized attacks (weight exfiltration, unmonitored rogue deployment, permission elevation, and more). A frontier monitor, GPT-5.6 Sol, reading each agent separately flags only 7% of successful distributed attacks, compared with 87% when a single agent carries out the same side task. Individual attacks can also be composed into longer attack sequences toward a greater loss-of-control outcome, without any individual agent being flagged. Finally, we tested whether models can plan these covert attacks: even with execution and monitor feedback, the best model-planned attacks succeed covertly against a single-trajectory GPT-5.6 Sol monitor in at most 25% of executions, compared with 80% for human-crafted plans, though this may scale with model capabilities. By turning worst-case coordinated attacks into reproducible red-team cases, our work provides a foundation for preventing locally benign agent actions from accumulating into system-level loss of control.

0.09

Cryptographically Secure Agent-to-Agent Communication in Multi-Agent Systems

cs.CR cs.MA ·2026-10-16 ·Karol Budzyn, Adrian Brodzik, Mateusz Szczepkowski
Abstract

Multi-agent systems (MAS) comprise AI agents that collaborate or compete to perform complex, distributed tasks. Their dynamic interactions create opportunities for threat actors to target and impersonate other agents, intercept messages, inject malicious prompts, and exfiltrate sensitive data while simultaneously complicating the detection of anomalous behavior through conventional monitoring. This paper surveys threats and cryptographic challenges in agent-to-agent communication and presents a zero-trust architecture integrating identity, transport security, message integrity, and monitoring. The architecture establishes agent identities via decentralized identifiers and task-scoped verifiable credentials backed by a permissioned ledger, enforcing least-privilege authorization. For transport security, it provides mutually authenticated, crypto-agile TLS with hybrid post-quantum key establishment, and evaluates KEMTLS with pre-distributed public keys to reduce handshake overhead. Signed CBOR envelopes preserve payload integrity across multi-hop exchanges, while zero-knowledge proofs verify declared task constraints without exposing underlying inputs. Context-aware monitoring of inter-agent messages, tool activity, and execution telemetry informs the setting of risk-based trust thresholds and supports the detection of behavioral anomalies. Together, these mechanisms provide an integrated framework for securing and auditing agent communication, with implications for other agentic protocols and prospective 6G semantic communication.

0.05

Systems & Networking 39

GUMPIM: Unitary and Malleable Memory for Processing-in-Memory with Guaranteed PIM Pages

cs.AR ·2026-10-16 ·Yilong Zhao, Fangxin Liu, Yiwei Hu, Zongwu Wang, Mingyu Gao +11 more
top venue
Abstract · 26 pages, 17 figures

DRAM-based Processing-in-Memory (PIM) addresses the "memory wall" by executing computations directly inside main memory. However, memory interleaving and virtual memory limit contiguous data size visible to PIM units, constraining PIM task granularity. To mitigate this, existing PIM systems drastically isolate PIM memory or disable memory interleaving. These design choices, however, decrease the CPU memory bandwidth and introduce extra data transfer, leading to an additional "system memory wall" that must be resolved to realize PIM's full potential. In this work, we propose GUMPIM, a PIM system that allows interleaved CPU pages and non-interleaved PIM pages to coexist in a Unitary and Malleable memory space with Guaranteed PIM page allocation. GUMPIM enables zero-copy during PIM task offloading and maintains CPU memory bandwidth while ensuring low PIM offloading overhead. First, we propose a dual-track memory management mechanism consisting of independent page allocation and address translation for CPU and PIM pages. Second, we design GUMPIM interface hardware on PIM-enabled DRAMs to provide a dynamic address mapping for the different data layouts of CPU and PIM pages. Third, we propose a PIM-assisted page migration mechanism that transparently migrates pages while preserving CPU access bandwidth, thereby enabling guaranteed and accelerated PIM page allocation. GUMPIM requires no changes to commodity DRAM standards; all hardware modifications are limited to the DRAM side, ensuring full compatibility with existing CPUs and enabling immediate deployment on current HBMx- and LPDDRx-based PIM platforms. Our results show only <0.1% performance degradation for CPU workloads on GUMPIM. For PIM workloads, GUMPIM reduces memory allocation and CPU-part computation times by 2.7x and 4.93x, respectively, yielding an end-to-end 2.3x speedup over a state-of-the-art baseline system.

0.71

Fluid-Spatiotemporal Stochastic Geometry for Information Flow in Non-Stationary Fields: Kinematics and Field Reconstruction

cs.NI ·2026-10-16 ·Wen-Yu Dong, Weiwei Jiang, Song Zhao, Qi Bi, Sheng Chen
published
Abstract · 18 pages, 8 figures. Accepted for publication in IEEE Internet of Things Journal

Classical stochastic geometry primarily addresses forward analysis by prescribing spatial point-process laws and mobility models. It does not directly address the inverse problem of recovering directional macroscopic transport from aggregate density evolution in non-stationary spatial networks. This paper develops the kinematic foundation of fluid-spatiotemporal stochastic geometry (F-STSG), representing an evolving mobile population by an intensity field under an assumed continuum limit while retaining the network infrastructure as a discrete spatial measure. We formulate macroscopic transport reconstruction as an inverse boundary value problem. Under a strictly positive bounded intensity and compatible source and boundary data, an instantaneous weighted minimum-energy criterion selects a unique velocity representative from the admissible class. The reconstruction reduces to a weighted elliptic Neumann problem and separates density-identifiable transport from weighted-solenoidal circulation that is not identifiable from density evolution alone. We further derive the information flux, interface transport, velocity-divergence relation, and centroid dynamics. Numerical experiments examine reconstruction consistency and sensitivity, while the public New Era of Urban Traffic Monitoring with Aerial Footage (pNEUMA) dataset provides held-out validation of internal transport. A separate unmanned aerial vehicle (UAV) study evaluates the centroid as a causal downstream control reference under observation delay, motion constraints, and execution perturbations.

0.63

qGAMMA: A Framework for Optimal DNN Mapping on Multi-Precision Accelerators

cs.AR cs.AI ·2026-10-16 ·Leandro Fiorin, Luigi Altamura, Cristina Silvano
published
Abstract · Accepted at the 25th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation (SAMOS XXV), 2025. To ap…

This work describes qGAMMA, a flexible framework for exploring and optimizing the mapping of mixed-precision Deep Neural Network (DNN) models onto general-purpose spatial accelerators that support a range of fixed- and floating-point precisions. qGAMMA navigates the design space using a domain-specific genetic algorithm and an analytical performance model based on hardware synthesis results. The framework identifies the optimal mapping and accelerator configuration that minimizes target metrics. Extensive evaluations across DNN workloads show that qGAMMA offers valuable insights for designing mixed-precision applications, enabling better trade-offs between performance, energy efficiency, and resource utilization.

0.62

LoopHIL: Looping Hardware into LLM-assisted Agents for Wireless Embedded Systems

cs.NI cs.AIcs.ARcs.SE ·2026-10-16 ·Dhairya Shah, Spanddhana Sara, Savitha Viswanadh Kandala, Ambuj Varshney
Abstract · 8 pages, 8 figures, 4 tables. Under review

LLM agents are widely used for software development and debugging, but their adoption in embedded systems, especially for wireless tasks, remains limited. This is because embedded development requires hardware-in-the-loop testing. Agents cannot directly observe the spectrum, probe pins, or measure power, so they cannot verify wireless link functionality; currently, human engineers perform this verification. We systematically evaluate whether LLM agents can establish wireless links when provided with hardware access. We introduce LoopHIL, a framework that lets agents control a multi-device RF bench, monitor it with a spectrum analyzer, logic analyzer, and energy meter, and validate links against a verification ladder, all within code-enforced safety limits. Across four increasingly complex tasks, from generating a carrier to backscatter decoding by a commercial receiver, and using six models from two vendors, hardware-in-the-loop increases task completion rates from 54% to 79% and from 17% to 83% on the most challenging task. It also halves task time and output tokens. With debugging tools, agents decrease tag power consumption by up to 96%. However, small models often report unverified claims, so independent hardware checks remain necessary.

0.56

SigmaTrace: Library-Agnostic Observability for CPU/GPU-Initiated RDMA Communication

cs.DC ·2026-10-16 ·Javid Baydamirli, Didem Unat
publishedcode/data
Abstract · Accepted at SC26. Author's accepted manuscript. 13 pages, 10 figures, 1 table. Code: https://github.com/ParCoreLab/SigmaTrace

RDMA networking enables low-latency, high-bandwidth communication on large systems. Despite its adoption, however, observability is lacking. Existing tools target specific communication libraries or only report aggregated data, missing per-operation behavior. The gap is widest for GPU-initiated networking, whose in-kernel RDMA operations are invisible to the host. We present SigmaTrace, a library-agnostic RDMA profiling framework providing fine-grained, per-operation visibility for CPU and GPU-initiated networking. It traces operations at the libibverbs/mlx5dv layers, the common denominator beneath communication libraries on InfiniBand and RoCE, and employs a low-overhead memory interposition technique that, to our knowledge, gives the first observability into GPU-initiated RDMA without source patching or binary instrumentation. Host-side tracing adds 1.3% overhead on NCCL AllReduce at 128 GPUs and under 2% on RCCL at 1024 GPUs. On a persistent-kernel BFS and Mixture-of-Experts inference, SigmaTrace exposes an in-kernel barrier spread and 2.13x inter-node traffic skew invisible to existing profilers.

0.52

HBFSim: A Simulator for Exploring HBF in Memory Hierarchies for LLMs

cs.AR cs.ET ·2026-10-16 ·Jinheng Li, Guosheng Wang, Hongchao Du, Shangyu Wu, Qiao Li +1 more
publishedcode/data
Abstract

High Bandwidth Flash (HBF) has recently emerged as a promising technology that provides terabyte-scale accelerator-local capacity at bandwidths approaching HBM, with a potential cost-per-bit advantage over HBM. However, its NAND foundation brings long latency, out-of-place updates, and finite write endurance directly into the accelerator memory hierarchy, making its effects on performance and lifetime difficult to predict. To study this emerging design space, we present HBFSim, an LLM-aware event-driven simulator for heterogeneous HBM+HBF memory systems that preserves application-level memory semantics while modeling flash-internal behavior. HBFSim models HBM, HBF with host-managed flash translation, and conventional capacity tiers in one simulation driven by a calibrated serving frontend. Using HBFSim, we study decoding of Llama-3.1-70B and Qwen3-235B-A22B under three workloads that differ in their input and output lengths. HBF serves data that does not fit in HBM 11.0--19.0$\times$ faster than host DRAM; HBM is better used to hold the running sequences' KV cache than as a cache for HBF; and at a published peak load of Llama-3.1-70B, HBF lasts 2.0 to 6.9 years, in proportion to the stacks that share the KV writes. The artifact is available at https://github.com/FlashLabAE/HBFSim and will track public HBF device data as it becomes available.

0.52

Joining Large String Collections with Jaro-Winkler Similarity

cs.DB ·2026-10-16 ·Manuel Lankes, Daniel Kocher, Daniel Schmitt, Willi Mann, Thomas Samy Dafir +3 more
top venue
Abstract · Accepted for publication in Proc. ACM Manag. Data (PACMMOD / SIGMOD 2027). Please cite the official version: https://doi.org/10.1145/3856354

The Jaro-Winkler measure, originally developed for U.S.\ Census record linkage, captures similarities between strings and accounts for common typographical variations. Despite its widespread use, finding all string pairs that exceed a given Jaro-Winkler threshold remains challenging as the string collection grows large. Existing algorithms for computing the Jaro-Winkler similarity between two strings are either slow for short strings or require quadratic time and fail to scale to long strings. To reduce the number of string comparisons, state-of-the-art techniques perform an index-nested-loop join on trie-based indexes; however, they have only been evaluated on small datasets and high similarity thresholds. For lower similarity thresholds that are relevant in practice, these techniques incur an excessive number of costly Jaro-Winkler computations. We present a novel approach for efficient Jaro-Winkler similarity joins on large string collections. Our solution is based on three key techniques: (1) JUMP is a fast algorithm for computing the Jaro-Winkler similarity between two strings that runs in linear time for bounded alphabets. (2) Dual-bound intersection bitmaps encode strings -- represented as character sets -- into fixed-size bitmaps, which are used to prune candidate string pairs by providing effective bounds on their character overlap; unlike prior bitmap techniques, our bitmaps are equipped with a cost model and adapt to the input data. (3) Our PUMA framework groups strings by prefix to tighten filter conditions, thereby reducing the number of similarity computations. Extensive experiments on large string collections demonstrate that our approach consistently outperforms the state of the art, achieving speedups of up to three orders of magnitude.

0.43

MailRail: Mitigating Incast Congestion with User-Space, Node-Based Flow Control

cs.DC ·2026-10-16 ·Noel Gomillion, Lance Fletcher, Trevor Steil, Eun Jung Kim, Roger Pearce
published
Abstract · To appear in IA3 2026. 9 pages, 10 figures

Irregular workloads, such as those stemming from Graph Neural Networks (GNNs) and Mixture-of-Experts (MoE) models, frequently induce severe incast congestion in large-scale parallel environments. Existing fabric-level congestion control mechanisms rely on reactive, link-level rate limiting that degrades overall throughput. While receiver-driven, credit-based flow control offers a proactive alternative, as process counts scale memory slots per rank fall below usable thresholds, rendering traditional rank-based accounting untenable. To address this challenge, we introduce MailRail, an out-of-order, asynchronous messaging framework, along with a novel user-space, node-based credit flow control system. By shifting credit accounting from individual processes to shared per-node pools, our proactive, fabric-agnostic approach maintains sufficient credit availability even as process density grows, enabling processes to dynamically utilize freed credits without waiting for receiver-side redistribution. Experimental evaluations demonstrate that node-based flow control achieves up to a 1132x speedup over unconstrained execution at an adversarial incast ratio of 4096:1 while adding minimal overhead on uniform workloads.

0.38

Merge Hulls Without Copying Vertices: GPU Isotonic Regression

cs.DC ·2026-10-16 ·Elena Raikova
theorycode/data
Abstract · 13 pages, 3 figures, 7 tables. v2: soft-rank, training-step and gradient results on all four GPUs. Code: https://github.com/LenaBarretta/tor…

Isotonic regression finds the monotone sequence closest to the data in weighted least squares. It is used to calibrate probabilities and inside differentiable ranking. Its standard algorithm, pool adjacent violators, is sequential within a sequence, and torchsort, the exact GPU solver of differentiable sorting libraries, runs it with one thread per sequence, which is slow on long ones. The fit is the slope sequence of the lower convex hull of cumulative sums, and hulls of sorted points can be merged in parallel. Merging partial hulls level by level, however, copies the surviving vertices at every level, which moves Theta(n log n) data when most points survive. We store each chunk's hull once and represent a merged hull as a list of segments of these chunk hulls, at most one per chunk; a merge then updates the list. We prove the representation correct, bound the work and depth of the merge phase, and measure a CUDA implementation on four NVIDIA GPUs against a control that copies the surviving vertices with the whole GPU at every level. On dense hulls the merge phase is 17 to 276 times faster with segment lists and the whole operator 3.9 to 19 times; on noisy data, where few vertices survive, the difference is small. In a differentiable soft rank, replacing torchsort's isotonic solver makes forward and backward 1.8 to 820 times faster, and the training step of a compact model with a Spearman loss 1.5 to 40 times. Code: https://github.com/LenaBarretta/torchisotonic

0.38

MAC units for a 2x2 int8 systolic array on Tiny Tapeout

cs.AR ·2026-10-16 ·Danila Gorodecky
code/data
Abstract

We take the multiply-accumulate (MAC) unit of an open-source 2$\times$2 int8 systolic array taped out on Tiny Tapeout (GF180 process) and propose two replacements. The first one, \emph{hybrid}, uses a different arithmetic structure for each row of the array. It reduces the logic area of one MAC unit by 4--6\% and of the whole chip by 0.7--3.2\% after place and route. The second one, \emph{fused\_eq}, never builds the long carry chain of the accumulation. It shortens the critical path by 20\% for one MAC unit and by 14\% for the whole chip. All numbers come from the standard Tiny Tapeout flow (LibreLane) run in GitHub Actions. Correctness is checked by exhaustive simulation of all $2^{24}$ inputs, a formal equivalence proof, and the original test bench.

0.37

The Rise of Personal Trackers as the Next Infrastructure for Ubiquitous Positioning

cs.NI ·2026-10-16 ·Mohamed I. Hany, Hamada Rizk, Moustafa Youssef
published
Abstract · 4 pages, 1 figure. Accepted at the 5th ACM SIGSPATIAL International Workshop on Spatial Big Data and AI for Industrial Applications (GeoIndu…

Reliable indoor positioning remains a central unsolved problem in spatial computing. Existing systems depend on fixed, provider-owned infrastructure (WiFi access points, Bluetooth beacons, UWB anchors) whose deployment, calibration, and ownership models limit their scalability. Meanwhile, personal trackers, wearables, smart luggage tags, and tagged industrial assets are becoming densely embedded in the very indoor environments where positioning infrastructure is weakest. This paper argues that these devices point to a new class of spatial infrastructure: non-stationary, user-owned, and semantically interpretable. We present TrackerNet, a vision for transforming personal trackers from private object-finders into a decentralized substrate for ubiquitous positioning. The key shift is to stop treating anchor mobility as noise: TrackerNet models mobile anchors as probabilistic references whose stability, trust, and semantic role are inferred over time, reframing indoor localization as non-stationary spatial graph inference. We identify three research frontiers (non-stationary SLAM over mobile, intermittently visible anchors; semantic reasoning that connects ranging signals to interpretations of anchor behavior; and consent-based, privacy-preserving governance for cross-vendor cooperation) and a staged pathway by which the vendors and venue operators who already run tracker networks can realize them with academia. We do not claim a finished system; we define a research agenda.

0.35

KVCompass: A Discovery Plane for Internet-Scale Prefix Cache Sharing in LLM Serving

cs.DC ·2026-10-16 ·Shiyi Ling, Fei Fang, Chen Qian
published
Abstract · 26 pages. Accepted for publication in Proceedings of the ACM on Measurement and Analysis of Computing Systems (POMACS)

KV cache reuse is one of the most effective ways to reduce LLM inference cost, but today its benefits are largely confined to a single cluster. When a provider serves the same model across geographically distributed regions, each region may repeatedly prefill the same popular prefixes, wasting GPU time and increasing time-to-first-token (TTFT). Exploiting reuse across regions involves discovering which worker currently holds a reusable KV cache for each request's prefix. We show that this KV cache discovery problem is structurally harder than conventional object lookup. Unlike hash-placement mechanisms such as DHTs, where an object's location is a deterministic function of its key, KV caches are created on the workers that serve requests, and their holder sets change with subsequent requests and KV cache evictions. KV cache discovery therefore calls for decoupling KV cache identity from placement. We present KVCompass, a discovery plane for Internet-scale KV cache sharing. The discovery index supports variable-length prefix lookup with constant-time local table reads at the edge, allowing each request to be forwarded to the worker that best balances KV cache reuse against load and network distance. Evaluation includes a 64-worker geo-distributed simulation and a 14-worker three-continent real-machine testbed. Evaluation shows that KVCompass achieves $1.3-14\times$ lower P99 TTFT than the baselines, up to $3.7\times$ lower P99 end-to-end latency, up to $4.7\times$ the throughput, and a peak load factor within 2% of the Oracle floor.

0.33

Surveying Experimentally-Validated Components for DNA-Based Molecular Communication

cs.ET ·2026-10-16 ·Andreas Waldner, Regine Wendt, Nils Stumpf, Sarah Scheer, Christian Hyttrek +2 more
surveyunusual
Abstract · 28 pages, 7 Tables, 8 Figures, under submission for IEEE COMST

Molecular communication is a novel paradigm for the design of nanonetworks. It uses molecules as information carriers and enables communication between nanodevices. DNA-based molecular communication is a promising approach for medical applications, such as in-body networks for diagnostics and drug delivery. However, the transition from theoretical models to practical experiments is still in its early stages. This survey reviews experimentally validated DNA-tiles, nanocarriers, and biosensors relevant to implementing DNA-tile-based nanonetworks in vitro. We evaluate these components based on their stability, compatibility, cost-effectiveness, and overall suitability for wet-lab implementation. Based on this analysis, we identify promising component combinations and derive a blueprint for a minimal proof-of-concept experiment. The resulting framework provides researchers with both a structured overview of experimentally relevant design choices and a concrete starting point for implementing and validating DNA-based nanonetworks in the wet lab.

0.31

An Error Taxonomy for Property Graph Data

cs.DB ·2026-10-16 ·Valerie Restat, Dominique Hausler, Désirée Bruns, Uta Störl
Abstract

Graph data are increasingly used in applications where relationships and topology matter. Yet, compared to tabular data, the notions of data quality and error types for graphs are less established. Tabular error concepts often fail to capture structural properties and context inherent in graph data, hindering systematic quality assessment in graph-based systems. We focus on property graphs (PGs) in graph databases. PGs are often managed without a prescriptive schema, enabling data insertion without explicit validation, while still exhibiting an implicit schema of recurring structural patterns. Irregularities in these patterns may indicate errors, but their interpretation depends on the surrounding graph context. To address these challenges, we provide a systematic view of errors in graph data by proposing a comprehensive graph-specific taxonomy of error types. To position these errors with respect to established work on tabular data quality, we derive a mapping between graph errors and corresponding error notions in tabular data. This mapping highlights which concepts transfer across data models and where structural properties of graphs introduce fundamentally different challenges requiring extensions. Overall, the taxonomy and mapping provide a shared vocabulary for describing, characterizing, comparing, and studying graph data errors, and they lay the groundwork for future work on error detection and repair in graph data.

0.29

An Embedded Industrial Irrigation System with Weather-Integrated Intelligent Scheduling

cs.ET ·2026-10-16 ·Kiarash Zangooei, Arya Masoodpay, Hamid Reza Ghaeini
Abstract · 9 pages

Efficient irrigation management requires timely responses to soil moisture conditions and anticipated weather changes. This article presents an embedded, weather-based irrigation system that integrates soil moisture monitoring using multiple sensors, edge-based decision-making, automated pump control, and local data management. Four capacitive soil moisture sensors are connected to an STM32F103C8T6 microcontroller for data acquisition, while a BTT Pi V1.2 platform evaluates the sensor data, retrieves short-term weather forecasts, makes irrigation decisions, manages a graphical user interface, and stores operational data locally. Individual sensor readings serve as the basis for zone-level irrigation decisions, while the median of four time-averaged readings supports monitoring and analysis. Forecasted precipitation is taken into account as effective precipitation to reduce the calculated irrigation requirement before the pump is activated. The system was evaluated as part of a one-month cultivation trial with radishes, during which a total of 38,112 data points from measurement cycles were recorded. The median value of normalized soil moisture averaged 47.93\% and remained predominantly within the configured control range. Three irrigation events with a planned total irrigation volume of 26.68 L were recorded. Finally, we demonstrate the technical feasibility of integrating sensors, weather data, local data processing, automated control, and data collection into a scalable platform for precision irrigation.

0.29

Data-Driven Prediction of Optimal MPI Rank Placement and Node Count for Memory-Bound CFD on AMD EPYC Clusters

cs.PF cs.LG ·2026-10-16 ·Marcin Lawenda, Łukasz Szustak, Aleksandra Krasicka
Abstract · 15 pages, 3 figures, 5 tables

Users who run a memory-bound OpenFOAM case on an AMD EPYC cluster must decide how many nodes to use and how many MPI ranks to place on each core complex. The fastest choice is often not the cheapest one. This paper revisits our earlier analytical model of this performance and energy trade-off with 560 measured configurations of the motorBike case on five Zen 4 and Zen 5 EPYC processors, including socket power from processor telemetry. On up to four nodes, small meshes are both fastest and cheapest on a single node on most processors. For the largest mesh the largest node count is both the fastest and the cheapest option, because its speedup is superlinear. A genuine trade-off arises only for medium meshes, where full speed costs a median of 20 to 97% more node hours than the cheapest option. The L3 miss rate and the halo traffic per cell are consistent with this pattern. For a new mesh of the same case, the configuration that was optimal on the nearest measured mesh is near optimal in 75 to 85% of the decisions. Running the two configurations it recommends for speed and for cost raises this share to 85 to 90% for runtime and node hours and determines the trade-off with median errors of 3 to 5 percentage points. Neither gradient boosted trees trained on the same data nor the earlier polynomial model improves on this simple procedure.

0.28

DIFFLET: A Fast Inference Engine for Diffusion Transformer Models on AWS Trainium

cs.DC cs.LG ·2026-10-16 ·Bin Ma, Min Yu, Vijay Kodamalla, Johnny Liu, Da Li +1 more
Abstract

On AWS Trainium2, every component of a diffusion model runs as a program compiled ahead of time for fixed shapes, while pipelines differ across models, prompt lengths across requests, and caching decisions across denoising steps. We present Difflet, a serving engine in which five model pipelines share one compilation backend, joint attention is split across cores by attention head, and a host controller decides at each denoising step whether to run the denoiser or reuse its earlier output. A small probe program computes the caching signal on the device and returns one number per step, and a resident model process keeps every program loaded across requests. Across five image and video model families, fused attention accelerates denoiser steps by 1.2-4.3x over the same framework using PyTorch/XLA attention, and at 32.8k visual tokens it is the only attention that compiles. On four of the five models, resident serving is 1.5-3.4x faster than standalone runs that reload weights from a warm file cache for each generation.

0.28

Who Plans, Who Edits? Verified, Cost-Aware Division of Labor for LLM-Driven Analog Layout Repair

cs.AR cs.AI ·2026-10-16 ·Xilai Dai, Saeed Mohammadi
Abstract · 11 pages, 8 figures, 6 tables. Submitted to IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems

Language-model agents can repair analog layouts when sign-off tools check every edit, but which model should plan a repair, which should edit it, and at what cost remain open questions. In our tool-grounded repair agent, models only choose bounded layout operators; sign-off tools rerun each task's contract checks after every edit, so failing layouts are never accepted. On layout-versus-schematic (LVS), design-rule checking (DRC), and post-layout repairs of three designs (a production high-voltage digital-to-analog converter (DAC), a 5-V oscillator, and a SKY130 operational transconductance amplifier (OTA)) with six tier-matched models from two vendors, 182 blind runs show that cheap models execute plans but rarely succeed alone (the cheapest model: 1 of 11 runs alone, 16 of 17 under a flagship planner; p < 0.001), cheap planners fail more often, and no fixed assignment is cheapest on every task. We propose a learning controller that selects models for each role and the escalation order from the first tool evaluation and updates Bayesian estimates after every attempt. With rules and initial estimates frozen before testing, the controller fixed 7 of 8 episodes on four sealed held-out tasks, matching the OpenAI flagship alone, at 15-40% of the flagship's spend per fix on three tasks but 8% higher spend overall. The cheapest mid-tier model alone fixed 6 of 8 at the lowest overall spend per fix but missed the hardest task. All 32 extracted DAC repairs match the human-guided LVS repair's integral nonlinearity (INL) within 0.005 least significant bits (LSB) at every code.

0.28

The Optimal Ranking Tradeoffs Between Two Ranking Scores in Two-Class Classification

cs.PF cs.CVcs.LG ·2026-10-16 ·Sébastien Piérard, Adrien Deliège, Marc Van Droogenbroeck
Abstract

The accuracy, true negative rate (negative class recall), true positive rate (positive class recall), negative predictive value (negative class precision), positive predictive value (positive class precision), class-specific intersections over unions, class-specific F-scores, Cohen's kappa, and balanced accuracy are scores that one intuitively wants to maximize when developing machine learning methods. However, they lead to different rankings of such methods. So, how can we establish a ranking that is an optimal tradeoff between any pair of these scores? It turns out that all these scores are special cases of ranking scores: a continuum of scores parameterized by the relative importance values assigned to the true negative, false positive, false negative, and true positive outcomes. Since all these scores are known to induce meaningful rankings, we seek the optimal tradeoffs among them. We consider the tradeoff between two scores to be optimal when it minimizes the Fréchet variance regarding the Kendall distance (also known as the bubble-sort distance) between rankings. For any given set of classifiers to rank, the optimal tradeoff lies at the center of the shortest (geodesic) path on the manifold of all rankings induced by the ranking scores. However, in the parameter space of the ranking scores, it is rarely achieved by averaging the importance values corresponding to the initial scores. In this paper, we present an exact analytical solution for determining a ranking score that is an optimal tradeoff for any given pair of initial scores, in the common case where the class priors are the same for all compared performances. We also experimentally investigate what happens when performance follows some distributions with fixed priors.

0.24

ConQUESTA: Topology-Aware Post-Training Structural Pruning of Quantum Neural Networks

cs.AR quant-pheess.SY ·2026-10-16 ·Saim Rehman, Muhammad Shafique
Abstract · Submitted to IEEE ISCAS 2027 for Review

Deploying quantum neural networks (QNNs) on sparse hardware requires preserving task accuracy while reducing compiler-induced two-qubit gates, SWAPs, and depth. We present ConQUESTA (\underline{C}onstrained \underline{Qu}antum \underline{E}ntangler \underline{S}election with \underline{T}opology \underline{A}wareness), a post training pruning framework that combines task saliency with compiler cost, screens candidate entanglers, and searches complete pruning masks under an explicit accuracy constraint. Unlike one-shot ranking, each mask is transpiled as a whole across multiple coupling topologies. Across 4-, 6-, and 8-qubit QNNs and eight model seeds at 55% pruning, ConQUESTA achieves mean resource reductions of 61.6% on Breast Cancer, 72.0% on Digits01, and 82.2% on Iris, exceeding the strongest matched simple baseline by 3.5, 1.7, and 8.9 percentage points, respectively. Fixed-mask recovery raises feasibility from 16/24 to 22/24 runs. In a fixed-k=4 scaling study on a scaling probe of 4 to 12-qubits, the raw mask space grew from 495 to 12.1M while the evaluated masks increase only from 189 to 305. Resource-weight changes alter depth/routing priorities with at most a 0.44-point mean validation shift.

0.24

One Communicator, Three Vendors: Extending the MoA Predictive Theory to a Real NVIDIA+AMD+CPU MPI Ensemble

cs.DC cs.LG ·2026-10-16 ·Lenore Mullin, Gaetan Hains
Abstract

The companion MPI-extension paper validated the first communication-bearing machine-shape ensemble: four NVIDIA A100 ranks and four CPU ranks on one NCSA Delta node under one Open MPI job, reassembled by MPI_Gatherv. This paper extends it to a multi-vendor communicator. A Slurm heterogeneous job spanning Delta's A100 and AMD MI100 partitions places A100, MI100 and CPU ranks (nine in all) under one communicator for the first time. We give the theory (a w_i proportional to 1/t_i weight split for three concurrent realizations; a het-job specification as an admissible directive) and report real results, including what the run exposed: the 1/t split assumes time linear in rows, and this fails for the MI100 rank (6 rows took 597 ms against a linear prediction of 1.6 ms). A launch-geometry model, with predictions recorded before measuring, succeeds for the A100 (time flat from 128 to 3456 rows, stepping at exactly 3520 and 6976) and places the MI100 floor in the key loop; changing its launch settings cut about 600 ms to 14.6 ms and the energy per call at 6 rows by 32x, though the MI100 remains about 95x slower than an A100 at 8192 rows. Controlled reruns show that the companion paper's 8.1x A100 result and 22 ms Gatherv were artifacts of four ranks sharing one GPU under Open MPI's default transport. A key/value-split partition whose message sizes are derived from shapes alone is then run: the ring Allreduce of the 8.5 MB message took 11.3, 16.0 and 15.3 ms against a predicted 11-12 ms, the small Allreduce missed its prediction in every run, and compute times reproduce to about 1% while communication times vary up to 4x between runs. Each idea is introduced with a figure and closed with worked examples.

0.24

LLM-Guided Dual-Stage Latency Optimization for Closed-Loop 6G V2I Digital Twins

cs.NI ·2026-10-16 ·Liu Cao, Peide Li, Yuan Yang, Zhaoyu Liu, Lyutianyang Zhang +2 more
Abstract · This work has been submitted to the IEEE for possible publication. 15 pages, 8 figures

Wireless Digital Twins are emerging as a key enabler for predictive network control in 6G Vehicle-to-Infrastructure (V2I) systems. However, closed-loop V2I DTs face two dual-stage communication bottlenecks arising from model parameter exchange for V2I DT reconstruction and multi-service V2I traffic delivery after each reconstruction. This paper proposes a closed-loop dual-stage End-to-End (E2E) latency optimization framework that integrates Compressed Personalized Federated Learning with a unified Large-Language-Model-guided Multi-Agent Reinforcement Learning approach. A CPFL-enabled decision-oriented V2I DT is first constructed by jointly predicting Path Loss (PL) and Quality-of-Experience (QoE)-oriented Key Performance Indicators (KPIs). Then the framework comprises two temporally coupled but latency-wise distinct optimization stages: Stage I minimizes the joint Uplink (UL) and Downlink (DL) E2E transmission latency of compressed shared backbone's parameters, and Stage II minimizes the service-aware E2E latency of multi-service V2I traffic. As the two stages share aligned agent roles and action space, we develop a unified Semantic-Aware Guidance-Enhanced Multi-Agent Proximal Policy Optimization framework, in which the LLM acts as a training-time semantic planner that provides high-level traffic splitting and power allocation guidance to reduce inefficient exploration, while the MARL agents execute feasible low-level actions. Simulation results demonstrate that the proposed framework enables efficient V2I DT reconstruction and significantly reduces both E2E latency for DT reconstruction and multi-service V2I traffic. The proposed SAGE-MAPPO supports cross-stage policy reuse by sequentially solving the two stages within each DT update cycle, which accelerates convergence relative to the existing baselines and facilitates timely DT updates in dynamic V2I scenarios.

0.23

PrairieFire: Igniting High-Throughput Updates in Large-Scale Approximate Nearest Neighbor Search

cs.DB ·2026-10-16 ·Yuchen Huang, Baiteng Ma, Yiping Sun, Xiao Chen, Rui Ma +7 more
Abstract

Arrays of modern NVMe SSDs are approaching memory in terms of bandwidth while offering much lower cost per gigabyte. Hence, approximate nearest neighbor search (ANNS) for billion-scale vectors in applications such as recommendation, search, and retrieval-augmented generation increasingly runs on SSD-based systems. Unfortunately, due to the limited update throughput of existing SSD-based solutions, the indexed corpus cannot keep pace with continuous modifications to data and embedding models, thus hindering broader deployment across more scenarios and larger scales. The bottleneck lies in CPU-intensive update paths, where foreground insertions incur frequent SSD accesses and background reassignments repeatedly relocate each vector, making high update throughput difficult to achieve even by exhausting available hardware. In this paper, using buffered updates and batched local rebuilds, we develop PrairieFire with cascaded adjustments, concurrency control, and CPU-GPU-SSD cooperation. These co-designs avoid frequent SSD accesses in foreground updates and heavy background reassignments, achieving high update throughput while maintaining excellent search performance. Evaluations show that, compared with state-of-the-art systems, PrairieFire can sustain approximately 3-35x update throughput while maintaining approximately 3-5x search throughput simultaneously. It also achieves approximately 1.5-8.7x cost efficiency. PrairieFire has been deployed in RedNote's production environments for two months, improving business metrics by higher freshness without frequent index rebuilds.

0.22

SemQBench: Benchmarking Decomposition-Aware Runtime Policies for Quantum-HPC Search

cs.ET ·2026-10-16 ·Chandrachud Pati, Yogesh Simmhan
published
Abstract · To Appear in the Proceedings of the 2nd International Workshop SFWM, co-located with SC26

Quantum processors are increasingly used as accelerators inside high-performance computing (HPC) workflows. A workflow runtime decides which quantum circuits to run, how to group them into provider jobs, and what to repeat after a failed attempt. Many search problems split into independent parts, but the runtime can use that structure only when the workflow exposes the parts and a validity check for each one. SemQBench measures what a runtime gains from this visibility. It compares seven execution policies for the same search problem, including two controls without amplitude amplification. The policies change one runtime choice at a time: whether to amplify, whether partial results may be combined across runs, whether the parts share one circuit, and what a retry keeps. We run them on the 20-qubit IQM Garnet processor using 106 distinct circuits in three jobs, and report circuit executions, active device time, and provider jobs separately before combining them in a cost model. Amplifying the problem as one whole circuit shows no statistically clear gain over random sampling and uses 6.7-104$\times$ more active device time. Exposing the parts and collecting valid partial results across a fixed batch of runs cuts the executions needed for 95% completion confidence by 2.20-4.00$\times$, and this policy is never worse than the other amplified policies in executions, device time, or jobs; it is selected in 99 of 108 overhead settings. Fewer executions do not always mean less device time: on one workload, executions fall from thirteen to four while active device time rises from 7.44 to 9.81 $μ$s. Exposing problem structure therefore gives the runtime a strong default policy and the measurements needed to decide when the simpler unamplified option is cheaper.

0.22

Encodings and Node Architecture: Lessons from the Fault-Tolerant Distributed Transversal Fanout

cs.DC quant-ph ·2026-10-16 ·Seng W. Loke
Abstract · 8 pages, 9 figures

An encoding choice specifies the quantum error-correcting code used to encode logical qubits at a given node. Different encoding choices may differ in the number of physical qubits, the number of encoded logical qubits, the code distance, and the available fault-tolerant operations. We study different encodings for a distributed transversal fanout operation and look at its implications on node architecture. Our experiments suggest that an encoding with higher distance, even if requiring more ebits, can yield a lower logical error rate for distributed operations, and that ``upgrading'' an encoding prior to distributed operations and having a cache, as part of node architecture, for qubits can be useful for distributed operations.

0.22

Preventing Data Hallucinations in Agentic AI for Optical Network Troubleshooting

cs.NI cs.AI ·2026-10-16 ·Jeremias Dötterl, Martin Kuipers
Abstract · Presented as a demo at the 52nd European Conference on Optical Communication (ECOC 2026)

An agentic AI system for automated troubleshooting in optical communication networks is demonstrated that integrates data references into the agent architecture to prevent data hallucinations, improving the interpretability and trustworthiness of root-cause identification and reasoning.

0.22

BaGuaLu-2: Taming Dynamic Memory Fluctuation for Efficient MoE Model Training over 10,000 GPUs

cs.DC ·2026-10-16 ·Mingshu Zhai, Minkun Guo, Hongrui Lian, Shuangyu Li, Qi Zhang +5 more
Abstract

Fine-grained Mixture-of-Experts (MoE) is the de facto architecture for scaling large language models. However, dynamic expert load imbalance causes severe memory fluctuations that degrade training efficiency. Existing systems rely on static memory provisioning: they either over-provision, leaving GPU memory idle during normal execution, or under- provision, inviting out-of-memory crashes and costly manual rollbacks when load spikes occur. To address these challenges, we propose BaGuaLu-2, a large-scale MoE training system that eliminates this dilemma through flexible provisioning. By transforming memory allocation from a static, one-time tuning choice into a continuous and fine-grained control. BaGuaLu-2 grants exact activation memory on demand for each training batch. Furthermore, it offsets any dynamic management overhead through system optimizations. BaGuaLu-2 demonstrates exceptional scalability, training a model with 1040 billion parameters across 12,288 GPUs while sustaining over 2 EFlops at 35% MFU. Compared with state-of-the-art systems, BaGuaLu-2 delivers 1.14x and 1.13x average throughput improvements over Megatron-Kuai and AdaPipe, respectively.

0.21

FuseRAG: Fusing Structure and Evidence into Inference for Cost-Efficient Graph-based RAG

cs.DB ·2026-10-16 ·Tingqi Wang, Yingshu Li, Zhipeng Cai
Abstract

Graph-based retrieval-augmented generation (RAG) exploits relationships among evidence to support multi-hop question answering. However, conveying retrieved graphs through non-unique linear textual contexts to Large Language Models (LLMs) leads to graph-structure degradation and inefficient KV-cache reuse. Meanwhile, graph-structure order optimization is already NP-hard, and the additional KV-cache reuse objective further complicates the search by introducing potentially conflicting orders. To tackle these issues, we propose FuseRAG, a cost-efficient graph-based RAG framework that shifts most retrieved evidence from explicit textual contexts to latent representations. Our main idea is to separate reusable evidence content from query-aware structural interactions in latent space, while retaining only a few explicit facts as semantic anchors. The resulting latent representations are injected into decoder self-attention, where they reshape attention over evidence and thereby influence hidden states and next-token probabilities. This design preserves graph structure through latent interactions, enables evidence reuse beyond exact token-prefix matching. Experiments on three multi-hop question-answering datasets with Llama-3-8B and Qwen3.5-9B show that FuseRAG improves answer quality while achieving up to 2.7x and 2.2x TTFT speedups over the fastest graph-based baseline, respectively.

0.21

BS-Spec: BFP-Based Bit-Sharing for Self-Speculative LLM Decoding with LUT-Driven Mixed-Precision Acceleration

cs.AR ·2026-10-16 ·Jiwoo Kim, Gunho Park, Sunwoo Yoo, Baeseong Park, Dongsoo Lee +1 more
Abstract

As large language models (LLMs) are increasingly deployed across diverse inference services, serving workloads span a wide range of batch sizes, context lengths, and performance requirements. Speculative decoding improves inference throughput, but conventional approaches require a separate draft model and KV cache, while existing self-speculative methods suffer from workload-insensitive draft configurations and inefficient mixed-precision execution. In this paper, we present BS-Spec, a block floating point (BFP)-based bit-sharing framework that jointly optimizes draft construction, workload-aware precision selection, and hardware execution. BS-Spec derives multiple draft representations directly from a single BFP target model by partially loading the most significant bits of weights and the KV cache while preserving shared exponents, eliminating separate draft-model storage. Offline calibration selects the weight and KV-cache precision pair that maximizes estimated speedup for each serving regime, and a precision lookup table applies the selected configuration at runtime. To support this adaptive execution, we design a mixed-precision BFP accelerator with a lookup table (LUT)-based variable-precision datapath for both draft and target computation. Experimental results show that BS-Spec achieves up to $1.9\times$ speedup over conventional autoregressive decoding and improves speedup over the evaluated bit-sharing self-speculative decoding baselines by $67.2\%$ on average, demonstrating the benefits of jointly co-designing bit sharing, precision adaptation, and specialized hardware for diverse LLM serving workloads.

0.20

ScaleRAN: A Runtime for Elastic GPU Resource Management in AI-RAN

cs.NI cs.DC ·2026-10-16 ·Arthur Hamard, Jérémie Leguay, Gopalasingham Aravinthan, Alberto Conte, Julien Lallet +1 more
Abstract · This work has been submitted to the IEEE for possible publication

AI-RAN is driving the convergence of radio access network (RAN) processing and artificial intelligence (AI) workloads on GPU-enabled computing platforms, ranging from GPU-accelerated base stations to cloud-native infrastructures. While GPU acceleration improves the efficiency of virtualized RAN functions, the computational demand of latency-critical baseband processing varies with traffic load, creating transient spare GPU capacity that can be reclaimed by co-located AI applications. Existing GPU-sharing mechanisms rely on coarse-grained hardware partitioning or static resource allocation, limiting their ability to adapt to slot-level workload fluctuations while preserving strict timing requirements. This paper presents ScaleRAN, a node-local runtime for elastic GPU resource management in AI-RAN. ScaleRAN leverages NVIDIA Green Contexts to dynamically partition Streaming Multiprocessors (SMs) between RAN and AI workloads through a ladder of pre-created GPU partitions, enabling low-overhead resource reallocation at RAN slot granularity. The proposed runtime naturally supports Open-RAN (O-RAN) O-Cloud deployments, one representative AI-RAN deployment scenario. To drive these reallocations, we propose a reactive controller inspired by Kubernetes autoscaling and a proactive controller based on quantile regression and conformal prediction. We also develop LDPCSim, a GPU-accelerated workload generator built on NVIDIA Aerial for reproducible evaluation. Results show that predictive fine-grained GPU partitioning satisfies slot-level RAN deadlines while increasing the GPU resources available to co-located AI workloads, demonstrating the practicality of elastic GPU sharing for future AI-RAN platforms.

0.20

NL2SQL with Iterative Self-Refinement

cs.DB cs.AI ·2026-10-16 ·Anjali Vemula, Akshat Dasula
Abstract

Bridging the gap between human intent and machine logic remains a critical challenge in data access for non-technical users, yet Large Language Models (LLMs) remain unreliable: they hallucinate schema elements, generate syntactically invalid queries, or fail to capture user intent. These limitations are especially critical in domains such as insurance, where KPIs, regulatory constraints, and legacy relational systems demand high precision. We introduce a multi-agent, persona-driven NL2SQL framework that combines (1) domain relevance detection, (2) schema-grounded knowledge construction, (3) KPI-aware SQL synthesis, (4) syntax-level self-repair using sqlite execution feedback, and (5) semantic consistency checking via SQL-to-NL reverse engineering. Our system significantly reduces query hallucination and improves semantic fidelity compared to direct LLM prompting. Evaluated on 60 queries (50 from a HuggingFace text-to-SQL dataset and 10 custom domain queries), the system achieves 90% correctness using Mistral across seven coordinated LLM calls showing that this approach can substantially increase the reliability of NL2SQL pipelines in high-stakes environments.

0.19

On the Power of Alternating Local Certification with Constant-Size Certificates

cs.DC ·2026-10-16 ·Pedro Montealegre, Martín Ríos-Wilson, Jukka Suomela, Guillaume Theyssier
Abstract

We study alternation hierarchies for distributed decision in anonymous networks with constant-size certificates. A prover and a disprover alternately assign certificates to nodes over a constant number of rounds, after which each node decides whether to accept based on its own input and the certificates observed in its neighborhood. An instance is accepted if all nodes accept. Unlike previous work on distributed alternation hierarchies, which considered restricted local computation (Reiter, PODC 2024) or logarithmic-size certificates (Feuilloley et al., ICALP 2016), we focus on constant-size certificates and the role of neighborhood information in determining expressive power. We consider three models: SET, where nodes observe only the set of certificate sequences of their neighbors; MULTISET, where multiplicities are also visible; and PORTS, where neighbors can additionally be distinguished by local port numbers. We establish that SET is strictly weaker than MULTISET, which is in turn strictly weaker than PORTS, and that all three alternation hierarchies are strict at every level. Despite their severe restrictions, these models are remarkably expressive: LEADER belongs to the third level of SET. Our main characterization shows that, over finite input alphabets, SET coincides exactly with monadic second-order logic (MSO$_1$). Moreover, every language definable in counting monadic second-order logic (CMSO$_1$) belongs to PORTS, while MULTISET already contains languages that are not CMSO$_1$-definable. Nevertheless, this expressive power has fundamental limitations. PORTS does not capture MSO$_2$: neither Hamiltonicity nor the existence of a perfect matching belongs to PORTS. Furthermore, over unbounded input alphabets, PORTS cannot verify that every node has been given the correct network size.

0.19

QuarkShard: Decomposing Cross-Shard Requests into Single-Input Transactions

cs.DC ·2026-10-16 ·Changyu Duan, Zhixuan Fang
benchmark
Abstract · 18 pages, 6 figures

Sharding blockchains enhance scalability by distributing the workload across multiple shards. Within sharding blockchains, securely and efficiently processing cross-shard user requests remains a pivotal challenge. Traditional sharding blockchain protocols execute cross-shard requests under a processing paradigm that treats each user request as a monolithic entity to preserve its atomicity, inherently introducing significant intra-shard BFT execution overhead and cross-shard communication overhead. We propose QuarkShard, a sharding protocol that breaks away from this traditional paradigm to achieve intra-shard and cross-shard overheads below the lower bounds imposed by conventional approaches. QuarkShard decomposes cross-shard requests into single-input-shard transactions, where each transaction requires consensus and processing within only a single shard, thereby substantially reducing total request processing overhead. Meanwhile, through a carefully designed request decomposition algorithm, QuarkShard guarantees fund atomicity or location atomicity during request execution. QuarkShard also enhances the security of blockchain sharding systems against MEV attacks and flooding attacks. Empirical evaluation across a wide-area network spanning four AWS regions demonstrates that QuarkShard achieves superior performance, attaining a peak throughput of $1024$k requests per second with sub-$4$-second latency. Benchmark comparisons show that QuarkShard achieves up to $210.7\%$ and $314.8\%$ of the maximum throughputs of Kronos and 2PC, respectively.

0.18

Understanding KV Residency in Production-Scale Agentic LLM Workloads

cs.DC ·2026-10-16 ·Mayank Arya, Agrim Bari, Anjaly Parayil, Yogesh Simmhan, Renee St. Amant +2 more
Abstract

Agentic large language model (LLM) applications turn one user request into a sequence of model and tool calls over a growing context. Retaining this context's key-value (KV) state in GPU memory lets later calls reuse it instead of recomputing it, but at fleet scale this needs substantial memory. We characterize four production Microsoft M365 workload traces (Deep Research, Enterprise Assistant, Coding Agent and Enterprise Chat) spanning more than 121 million model calls. In Deep Research, the application issues 89% of model calls and the median context grows from about 10K to 68K tokens by the 30th call. About 80% of context tokens are served from the prefix cache. With the public Llama-2-70B KV layout standing in for confidential models, one 99th-percentile Deep Research session exceeds the 80 GB HBM of an NVIDIA H100, and holding all live sessions at their peak needs an estimated 501.8 TB of HBM, although inference occupies only 5.6% of that retention time. Trace-driven replay shows that a one-hour idle cap cuts this ceiling by 3.6x while raising the share of calls that recompute context only from 4.07% to 4.53%. Under tighter budgets, eviction alone is insufficient: at one tenth of the busiest hour's demand, LRU eviction makes 63% of Deep Research calls recompute and clairvoyant furthest-next-use 31%, whereas multi-resolution retention (MRR), which demotes the LRU session's KV state to int4, lowers this to 13.6%. Placement imposes a separate limit: when the next call reaches a different deployment, cached-context reuse drops from 89% to 29%, and treating KV state as deployment-local raises the recompute floor under unlimited memory from 5.4% to 20.2%. These findings motivate retention that changes the representation of KV state, together with deployment-aware routing or migration. We plan to release AgenticTraceSim, a parametric trace generator, and a residency replay framework.

0.16

TileDRAM: Compiler-Controlled Tile-to-DRAM Layouts for 3D DRAM GPGPUs

cs.AR ·2026-10-16 ·Guangyuan Xu, Lanxuan Zhang, Ziyang Li, Zhaoshi Li, Yajun Ha
Abstract

Modern GPU compilers expose rich layout abstractions for mapping tensors to threads and on-chip resources, but compute-unit-local (CU-local) 3D DRAM architectures lack a corresponding abstraction for specifying how operand-tile data are arranged within the DRAM address space. Without such an abstraction, conventional placement under the existing DRAM address mapping can subject structured tile accesses to severe row-buffer conflicts. We present TileDRAM, a compiler-runtime toolchain that introduces tile-to-DRAM layouts as a compiler-visible abstraction for DRAM placement. TileDRAM extends the GPU Linear Layout abstraction from compute-side mapping targets to DRAM placement coordinates and selects layouts per tile access signature, using recurring tile-local access behavior as the specialization unit. Since row-buffer locality depends on ordered request history, TileDRAM evaluates candidate layouts through their row-buffer behavior under fixed DRAM address mappings. Accesses under the selected layouts are realized through ordinary address arithmetic, without new address-remapping hardware or ISA support. In full-trace replay on Accel-Sim/Ramulator2 across 108 mixture-of-experts (MoE) feed-forward network (FFN) cases, TileDRAM reduces the request-weighted row-conflict rate from 75.7% to 15.2% and achieves a 1.77x geomean speedup. Retargeting six traces to six 3D DRAM organizations yields a 2.20x geomean speedup under the same compiler abstraction. On an RTX 3090, address transformation adds 0.17% geomean overhead over identity.

0.15

Federated Multiagent Soft Actor Critic for IRS Assisted Cognitive Radio Networks under Partial Observability

cs.NI ·2026-10-16 ·Satvika Pullisani, Usama Mir, Ubaid Abbasi
Abstract

Intelligent reflecting surfaces (IRSs) can improve spectrum efficiency and interference control in cognitive radio networks (CRNs) by reconfiguring the propagation environment. This paper presents FMASAC, a federated multi-agent soft actor-critic framework for IRS-assisted CRNs under partial observability, in which distributed secondary users (SUs) jointly select their transmit power and the phase shifts of an assigned subset of IRS elements using only local observations. Primary user (PU) protection is incorporated through an interference-penalized reward, and SUs coordinate by periodically averaging model parameters at a federated server, without exchanging raw observations, channel state information, or replay data. We formulate the problem as a partially observable Markov decision process, specify the local learning and aggregation procedure, and characterize its per-round communication cost. Illustrative performance trends contrast FMASAC with centralized SAC, independent SAC, IRS-fixed multi-agent reinforcement learning, and a heuristic CRN strategy. We also outline the simulation-based evaluation and the open challenges, namely constraint handling, agent heterogeneity, and scalability, that are needed to validate the framework.

0.13

Fast BSS Transition for Multi-PSK Community Wi-Fi in MDU Networks: Fragmented RADIUS Key Delivery, Server-Derived PMK-R0/PMK-R1 Distribution through a Cloud Management API, and Lease-Driven Key Revocation

cs.NI ·2026-10-16 ·Azif Abdul Salam Kolisam Beevi, Anil Kumar Reddy
Abstract · 30 Pages, 8 Figures, 20 tables

Multi-dwelling-unit (MDU) properties increasingly deploy one community-wide SSID on which every unit or resident has its own pre-shared key, mapped to its own VLAN so that each unit is a separate network. Vendors name this capability differently (ePSK, iPSK, MPSK, DPSK, PPSK); we call it multi-PSK (mPSK). When the keys are held in an external RADIUS server or a cloud controller, the access point (AP) must ask the key server to identify a client's key. Enabling IEEE 802.11r Fast BSS Transition with WPA2-PSK (FT-PSK), which large properties need for roaming voice and video, breaks this exchange in three ways. First, the server must identify the key from an FT handshake whose MIC depends on the FT key hierarchy. Second, the RADIUS messages no longer fit: EAPOL-Key message 2 and the key material the AP needs both exceed the 247-octet value capacity of a RADIUS vendor-specific attribute (VSA). Third, the AP a device roams to needs a PMK-R1 for that device, and all key material must disappear when the resident leaves. We present FT-mPSK. The key server identifies the PSK, derives PMK-R0 and PMK-R1, and returns the serving AP's keys in a two-fragment Key-Bundle VSA that the AP reassembles and authenticates before decryption. It then calls one POST API on the cloud management system, which pushes PMK-R0 to the primary AP. On a roam, the target AP obtains PMK-R1 from the primary AP, located by the R0KH-ID in the FT Authentication Request, or from the key server. When the Property Management System reports a move-out or lease end, the key server issues the same POST as a delete keyed by the resident's mPSK username, and every AP purges that username's PSK, PMK, PMK-R0 and PMK-R1. We specify the external-RADIUS and cloud-controller methods, the attribute and fragment formats, the API, algorithms, size budgets, complexity, security properties and the validation plan.

0.13

PairSight: Characterizing and Predicting Operator-Level Interference in GPU Sharing for DNN Inference

cs.PF ·2026-10-16 ·Xiao Liu, Yujia Fu, Yue Yu, Dan Huang
Abstract · 5 pages, 4 figures, 2 tables

Serving DNN inference is costly, yet a single task seldom saturates both the compute units and the memory bandwidth of a GPU. GPU sharing raises utilization by co-locating several tasks on one GPU, at the cost of interference among them. Existing systems estimate this interference with binary compute-memory rules, standalone scaling curves, or model-level predictors. However, these estimates rest on a limited understanding of interference. They leave the software and hardware properties of co-located operators largely unexplored and give no slowdown prediction that is both interpretable and accurate. We conduct an operator-level study of interference under SM partitioning and use hardware counters to trace it to contention on the shared memory path, where the slowdown depends jointly on both co-located operators. Based on these findings, we propose PairSight, a hardware-aware interference predictor that learns how the hardware behaviors of co-located tasks interact and predicts the latency of each task before co-location. On six DNN models including LLMs, PairSight predicts latency under sharing with 4.3% mean error, against 19.0% for a compute-memory intensity rule, and adapts to unseen models and inputs with a few samples.

0.10

Agentic AI for Heterogeneous System Design Across Chip, Scale-Up, and Scale-Out Domains

cs.AR cs.DC ·2026-10-16 ·Murali Emani, Kashyap R Manjusha
Abstract

The design space of modern AI/HPC systems has expanded rapidly across compute architectures, including GPUs, dataflow processors, and RISC-V accelerators; memory tech- nologies, including HBM, GDDR, and SRAM; and interconnects such as NVLink, UALink, Infinity Fabric, InfiniBand, Slingshot, and Ultra Ethernet. No single architecture is optimal across LLM training, LLM serving, and AI-coupled HPC workloads, while the resulting combinatorial design space is too large for manual exploration. We present ArchAgent, a hierarchical multi- agent framework in which an LLM-based planner decomposes system design into chip, scale-up, and scale-out subproblems, each handled by specialist agents. A deterministic analytical evaluator closes the loop, ensuring that all performance estimates are derived from analytical models rather than generated by the LLM. On a mixed AI+HPC workload under a 1 MW power budget, ArchAgent discovers heterogeneous designs that achieve a 1.93x weighted geometric mean speedup over a homogeneous GPU baseline, while requiring only 18 evaluations compared with 100 evaluations for hill climbing, which achieves 1.65x speedup.

0.08

Software & Programming 28

SpinControl: Control Software for Prototyping and Benchmarking of Spin-Qubit Devices and Measurement Protocols

cs.SE quant-ph ·2026-10-16 ·Harold B. Meerwaldt, Daniël P. van der Velde, Pieter T. Eendebak, Michiel J. Haye
code/data
Abstract · 11 pages, 8 figures

Quantum computers require software to control the states of the qubits. As the actions on qubits grow in complexity with circuit execution, calibration, and benchmarking, the complexity of the control software grows in tandem. Spin-qubit quantum computers are at a stage where experimentalists need low-level access for tuning and calibration, and quantum algorithm researchers want high-level circuit-level access. In this article, we present the software design of SpinControl. SpinControl integrates three architectural metrics: 1) customizability, to allow prototyping devices and measurement protocols, is achieved by separating data and behaviour; 2) modularity and extensibility are achieved by separating functionalities into different components; and 3) robustness, to reliably execute quantum circuits on many qubits, is achieved by CI pipelines and input validation. SpinControl allows different types of measurements while customizing the instruments used, the pulse or circuit program executed, and the data processing. It allows building standardized experiments from measurement and analysis building blocks and executing these for calibration in a custom and conditional order. SpinControl implements these functionalities alongside robust configuration and data management, a connection to the cloud, and calibration scheduling. Pulse and circuit compilation, virtual gate matrices, as well as spin-qubit-specific GUIs integrated by SpinControl, make heavy use of software packages developed by Delft University of Technology: pulse lib, core-tools (both open source), and QConstruct

0.41

No One-Size-Fits-All: Adaptive Code Editing with Feature-Based Strategy Selection

cs.SE ·2026-10-16 ·Jun Wan, Zhongxin Liu, Dajun Chen, Wei Jiang, Yong Li +1 more
top venue
Abstract · 23 pages, accepted to ISSTA 2026

Large language models have shown strong potential in automating code editing (CE) tasks, yet most existing systems rely on a fixed CE strategy (i.e., one-step, search-replace, and plan-apply) across diverse scenarios. In practice, CE tasks vary widely in code context, edit locality, and semantic complexity, causing no single strategy to perform consistently well. Moreover, systematic studies on the relationship between strategies and CE tasks remain scarce, limiting further optimization of LLM-based code editing strategies. To address this gap, we construct CEBench, comprising 500 real-world CE tasks curated from SWE-bench. Then we compare three representative CE strategies (one-step, search-replace, plan-apply). Our study reveals two key findings: (1) each strategy exhibits distinct advantages for specific CE tasks, with no universal optimality; (2) strategy effectiveness is significantly influenced by four core task features: total lines of code, modification type, modification scope, and task type. Motivated by these findings, we propose Ace, an adaptive code editing framework that dynamically selects optimal CE strategies based on task features. Unlike classifier-based or instruction-prompting approaches, Ace employs a fine-tuned LLM that learns feature-strategy mappings from empirical data to recommend strategies with interpretable rationales. Extensive experiments on our CEBench-verified show that Ace outperforms state-of-the-art baselines: it boosts Pass@1 by 21.7% while reducing token consumption by 27.3% compared to the best baseline (AutoCodeRover). Our strategy selector achieves 72.6% selection accuracy with 69.9% rationale accuracy (Cohen's Kappa = 0.807), providing transparent and trustworthy decision-making for developers.

0.38

VERAS: Verification-Guided Adaptive Search for Reliable Operations Research

cs.SE cs.AI ·2026-10-16 ·Minbo Ma, Honglin Zhang, Bangtao Zhou, Chunguang Yu, Dasheng Ju +3 more
code/data
Abstract

Advances in reasoning, coding, and tool use enable LLM agents to solve operations research (OR) problems through iterative problem modeling, program construction, and verification. Yet passing self-authored verification does not ensure that the solutions satisfy task constraints and meet optimization quality. This gap can arise from correlated specification errors, where the generated program and its checks share omissions or misinterpretations of task requirements, alongside incomplete verification coverage and ineffective optimization strategies. We propose VERAS, a verification-guided adaptive search harness for reliable OR that uses verification evidence to prioritize feasibility repair, constraint-coverage review, and solution-quality improvement. VERAS combines task-specific solution checkers with a separate verification agent to examine candidate solutions against the task requirements. It uses verification feedback to guide feasibility repair, additional constraint checking, and objective improvement, while candidate acceptance criteria screen revisions for deterioration in output validity or constraint satisfaction. On ORAgentBench, VERAS achieves a 42.99% pass rate, exceeding the reported 35.51% of Codex using the same LLM. Across two industrial-scale OR problems, VERAS reduces inter-datacenter peak bandwidth by 10.34% relative to native Codex and lowers annualized distribution network planning costs by 6.28% compared with expert-designed solutions. Code and evidence are available at https://github.com/holimario/VERAS.

0.37

Code Skill Server: A Large-Scale Search Engine for Code Generation Skills

cs.SE cs.IR ·2026-10-16 ·Xiaoyong Li, Haolong Li, Dongliang Xu, Haoyu Li, Yifan Wang +5 more
code/data
Abstract

Agent skills have proven to be an effective technique for improving agent ability on specific tasks, including software engineering, natural science, and mathematics. During such a process, using agent skills often involves manually discovering and organizing skills from heterogeneous sources, while existing skill retrieval methods mainly focus on selecting relevant skills from established skill libraries. However, with the rapidly increasing number and variety of reusable agent skills, manually searching for relevant skills across heterogeneous sources is becoming impractical at scale. In this paper, we introduce the Code Skill Server (CSS), a large-scale search engine for finding the most useful code generation skills for a target task. CSS consists of a skill server that incrementally collects and indexes publicly available code generation skills from multiple sources, with 993 skills included in the corpus used in this work. Extensive experiments across different retrieval methods and code generation benchmarks show that the best CSS configuration improves Pass@1 by up to 13.91% and consistently improves all evaluated models by at least 4.21% in the cross-model evaluation. These results demonstrate that CSS can effectively search for and serve relevant skills, improving code generation quality and enhancing models' code generation capabilities. Code and the CSS-993 corpus are available at {https://github.com/haoyuli-sdu/CodeSkillServer}.

0.37

QRecoupling.jl: numerical and exact computation of classical and quantum recoupling coefficients

cs.MS gr-qcmath-phphysics.comp-ph ·2026-10-16 ·Seth K. Asante
code/data
Abstract · 21 pages, 4 figures, 5 tables. Software: https://github.com/sethkasante/QRecoupling.jl (v0.6.4, doi:10.5281/zenodo.24027544)

The Julia package QRecoupling evaluates classical and quantum recoupling coefficients, such as Clebsch-Gordan coefficients and $3j$ and $6j$ symbols, and user-defined sums of factorial products, with numerical, exact and symbolic targets. One compact factorial rule covers classical and deformed evaluations, while its cyclotomic exponents reveal common factors and determine which terms contribute at unitary levels. Adaptive arithmetic resolves strongly cancelling sums, while exact tests in quantum integers distinguish algebraic zeros from small nonzero values without a computer-algebra system. At spin cutoff 500, classical $3j$ and $6j$ symbols are evaluated about 7 and 15 times faster than with WIGXJPF. For the tested quantum deformations, relative errors remain below $4\times10^{-14}$ through spin cutoff 400. Recurrences compute fusion matrices with 401 channels about 100 times faster than loops over their entries. The resulting coefficients support recoupling identities, changes of coupling basis and tensor-network calculations.

0.36

MLOps-Bench: Benchmarking AI Agents on Production ML Engineering in Real Repositories

cs.SE ·2026-10-16 ·Quoc Co Tran, Hitesh Laxmichand Patel, Diwakar Mahajan, Kshitij Bakliwal, Avi Sil +1 more
top venuebenchmark
Abstract

AI agents are increasingly expected to modify and maintain machine-learning systems in production, yet existing benchmarks primarily evaluate either competition-style ML experimentation or generic repository issue resolution. We introduce MLOps-Bench, a repository-grounded benchmark for production ML engineering. Unlike prior ML-agent benchmarks, MLOps-Bench evaluates executable lifecycle changes inside real ML repositories, spanning data pipelines, feature engineering, model training, model serving, monitoring, CI/CD and governance, and end-to-end integration. Each task is grounded in repository evidence and includes a natural-language requirement, oracle patch, plausible buggy baseline, fail-to-pass tests, pass-to-pass tests, and provenance metadata. The benchmark contains 212 validated tasks from 31 public ML repositories, with 636 fail-to-pass tests and 856 pass-to-pass tests. We evaluate six agents and find that the best system solves 31.2% of tasks while passing 68.7% of tests, showing that current agents often make partial progress but still fail production-relevant behavioral requirements. MLOps-Bench shifts agent evaluation beyond model building and issue fixing toward repository-native MLOps work where correctness depends on lifecycle reasoning, behavioral robustness, and preservation of existing system behavior.

0.36

Pattern-Based Precondition Inference from Method Implementations

cs.SE ·2026-10-16 ·Shanyue Zhou, Martin P. Robillard
code/datahot direction
Abstract

Method preconditions define the conditions that must be satisfied before a method is called. Often, such conditions are not documented and must be inferred. We present Prex, a pattern-based static analysis approach that infers preconditions from explicit checks in Java source code. For each precondition, Prex reports the condition required for a valid call and the reported exception in the case of a violation. Prex also attempts code transformations to convert the condition detected within a method's implementation into an expression that can be evaluated by client code. It leverages the concept of a mutation boundary to separate precondition checks from exceptions that may occur after the method changes the program state. We evaluated Prex from 46 open source Java projects and manually reviewed independent samples of 150 reports. Among the 150 Prex reports, 149 were correct and 125 of the correct reports were executable by client code. In a systematic comparison with an alternative approach based on path expansion, Prex provided more precise results. In total, 122 of the 149 correct reports had no relevant method documentation, and 41 required nontrivial source code analysis to recover manually. Overall, explicit source-code checks provide precondition information that can be absent from method documentation, and Prex can help recover it.

0.35

Towards an Intelligent Decision Engine for Smart Contract Deployment

cs.SE cs.AIcs.CR ·2026-10-16 ·Oana-Otilia Captarencu, Emanuel Onica, Andrei Arusoaie
publishedunusual
Abstract · Accepted for the 30th IEEE International Conference on System Theory, Control and Computing, ICSTCC 2026; Copyright is with IEEE

Layer 2 (L2) blockchain solutions emerged in recent years, predominantly in the Ethereum ecosystem, as a viable option for scaling a main blockchain network (L1). However, the selection of a L2 solution for deploying smart contracts backing blockchain applications is a decision that depends on many factors, such as costs, efficiency, popularity, and stability. We propose here a systematic methodology to help with this decision making process by using a set of relevant metrics and introducing user-defined profiles that capture different priorities among these metrics. We collect the major categories of smart contracts from a community-maintained data source, and we gather transactions for a six months period from three major L2 solutions: Arbitrum, Optimism, and Base. For each contract category and each blockchain, we compute a set of metrics and scores tailored to each user profile. By applying our methodology, we identify a clear bias in the classification, showing which L2 solutions are better suited for particular application types and user profiles.

0.32

Controlled PROP: Theory and Applications to Quantum Circuits

cs.LO quant-ph ·2026-10-16 ·Noé Delorme, Simon Perdrix
Abstract

Control is a fundamental concept in quantum and reversible computational models. It enables the conditional application of a transformation to a system, depending on the state of another system. We introduce a general framework for diagrammatic reasoning featuring control as a constructor. To this end, we provide an elementary axiomatization of control functors, extending the standard formalism of PROPs to controlled PROPs. As an application, we show that controlled PROPs facilitate diagrammatic reasoning for quantum circuits. In particular, they admit a simple and complete axiomatization involving at most three qubits, whereas in the standard PROP setting, completeness necessarily requires axioms involving arbitrarily many qubits.

0.28

SWE-Bench Pro V2: A Cleaner, Harder-to-Game Coding Benchmark

cs.SE cs.AI ·2026-10-16 ·Soham Dan, Jeff Da, Sanyam Bhutani, Miguel Romero Calvo, Ivan Guevara +6 more
benchmarkcode/data
Abstract · Dataset: https://huggingface.co/datasets/ScaleAI/SWE-bench_Pro/viewer

SWE-Bench Pro is a widely used benchmark of long-horizon software engineering tasks for coding agents. As agents improve, they reach task issues earlier agents never touched and interact with the evaluation environment in ways the original harness did not anticipate. We present SWE-Bench Pro V2, a revised public split built to keep the benchmark accurate as capabilities advance. Agentic audits, expert adjudication and blind re-implementation repaired or retired 480 of the 731 tasks: 23 contracted engineers spent 404 hours, each working in a language they write professionally, and every revised instruction was verified by a second engineer solving the task from the text alone. A hardened protocol restricts network access to the model endpoint and re-grades every patch in a clean container, so agents cannot retrieve upstream fixes or get credit for changes outside their patch. On the 642 V2 tasks frontier agents resolve up to 99.4% while other systems stay clearly separated, and a 51-task Hard split spreads systems from 52.9% to 98.0%. We release the benchmark and distill lessons for maintaining agent benchmarks over time.

0.27

Does AI-Assisted Development Homogenize Software? An Empirical Study of Code Similarity, Quality, and Evolution

cs.SE cs.AI ·2026-10-16 ·Luciano Marchezan, Romain Lefeuvre, Kevin Delcourt, Benoit Combemale, Houari Sahraoui
code/data
Abstract

AI-assisted software development is evolving from interactive LLM assistance toward agentic systems that autonomously perform software engineering tasks, yet the effects of this shift remain insufficiently understood. We present an empirical study of 50 long-lived Java and Python open-source projects across three eras: pre-LLM, potentially LLM-assisted, and detected agentic development. Using snapshots from Software Heritage, we examine code similarity, software quality, and software evolution. Similarity is measured semantically through embedding-based cosine similarity and syntactically through CodeBLEU*. Quality is assessed with SonarQube metrics covering maintainability, complexity, reliability, and security. Evolution is characterized by annualized growth rates of LOC, functions, classes, files, and statements, and by commit activity in the three months preceding each snapshot. Results show no consistent evidence of increased similarity as semantic and syntactic similarity decreased significantly in Java for both the LLM and agentic eras relative to the pre-LLM era, while Python showed no significant differences. Software quality remained largely stable, although the agentic era showed significantly lower code-smell and SQALE index densities. Structural growth was generally lower in the agentic era, particularly for files, whereas development activity increased moderately in Java, especially in commits and added lines. These findings challenge the assumption that LLM- and agent-assisted development necessarily homogenizes code, degrades quality, or accelerates growth.

0.27

Finitely Verifiable Segmentation: Decidability for Every Tokenizer with a Finite-State Verifier

cs.FL cs.CLcs.PL ·2026-10-16 ·Nicklas Nidhögg
theory
Abstract · 39 pages, 1 figure, 2 tables. v2: revises the introduction to credit existing exact procedures (Koo et al. for masks, Berglund et al. for BP…

A segmentation policy assigns to each byte string at most one segmentation into tokens. Call it finitely verifiable when a finite automaton over marked strings, bytes paired with boundary bits, accepts exactly the policy's own segmentations. The automaton judges a proposed segmentation and need not compute one. From a supplied verifier the paper gives constructive decisions for four questions, whether a window certifies a token boundary, whether the boundary gap is bounded, whether a mask can keep generation from dead-ending, and whether two policies are equivalent. The last three specialize limitedness, co-accessibility and the equivalence of single-valued transducers. The first refines the synchronizing cut of code theory, and two more ask whether a certified cut, or the cuts of an inventory, let each piece scan alone. The class sits strictly between the online-scannable policies and those with regular domains, and online scannability is decidable in polynomial time through Choffrut's theorem. Markings are length-preserving, so the class is exactly the rational segmentation policies, a normal form rather than a restriction. The class is closed under four pipeline operations, composition over a labeled stage, restriction to a regular schema, splitting at character-class changes, and byte-fallback union of literal policies. Each composite inherits the six decisions, and certificates transfer between stages under four laws. Under rational, unifilar, almost surely stopping finite-state sources, each mass of the weighted section is an exact rational, and so is each rate where expected length is positive. Maximal-munch lexing, canonical byte-pair encoding and two-mode WordPiece instantiate the interface, and the munch library implements all six decisions. Every theorem carries an executable check, and every reported statistic is a program's own output.

0.24

Model Checking for Noisy Toric Words

cs.FL cs.LO ·2026-10-16 ·Mihir Vahanwala
theoryunusual
Abstract

Toric words have recently gained attention as a means to model the behaviour of dynamical systems such as linear loops. This work is concerned with describing when the toric model is precise enough to decide whether the trace of a system is contained in an omega-regular language. Our contributions are threefold. (i) We use topological means to generalise a class of toric words by defining noise-robust dynamical systems and their sets of noisy traces. We prove that all noisy traces of a system are almost-periodic, and have the same set of recurrent factors. (ii) We apply the abstract techniques above to concrete sequences obtained as sign descriptions of real algebraic linear recurrence sequences (LRS), and show that the language-membership problem for sign descriptions of LRS with few dominant roots is decidable. (iii) We show that noisy traces of noise-robust dynamical systems are indistinguishable to a prefix-independent omega-regular language, and inclusion is decidable provided the common language of their recurrent factors is recursive. We incidentally obtain properties of prefix-independent omega-regular languages that may be of independent interest.

0.23

LLM-Assisted Bi-Approximate CHC Solving

cs.LO cs.AI ·2026-10-16 ·Baiting Jiang
Abstract · 22 pages, 8 figures, 5 tables

Constrained Horn clause (CHC) solvers often struggle to discover deep counterexamples whose derivations may require extensive rule unfolding. Although large language models (LLMs) can propose long-range symbolic relations, prior work on formally checked LLM-assisted verification has mainly focused on generating inductive invariants and concrete traces. We exploit this broader capability through Derivation Power Under-approximations (DPUs). Each DPU is a solver-certified under-approximation of the relation induced by a multi-step CHC derivation. An LLM proposes each DPU with symbolic intermediate-state witnesses that enable SMT validation without introducing additional quantifiers. Validated DPUs can be composed and reused soundly. Building on this mechanism, our LLM-assisted framework HornCraft combines invariant generation for over-approximation with DPU generation for under-approximation in CHC solving. On 260 linear and nonlinear integer-arithmetic CHC-COMP benchmarks left unresolved by a 30-minute Spacer run, HornCraft returns certified verdicts for 228 instances. It certifies 42 instances as unsafe, compared with 22 when DPU construction and reuse are disabled.

0.23

Symmetric two-counter machines

cs.FL ·2026-10-16 ·Krzysztof Jaros, Łukasz Kamiński, Maja Kądziołka, Sławomir Lasota
theory
Abstract

Counter machines are a particularly simple, Turing-complete model of computation. We study the expressive power of nondeterministic 2-counter machines which are symmetric, in that their set of update transitions is invariant under swapping of counters. Our model is an extension of symmetric 2-VASS, in that we allow modifying both counters within a single transition. As the main result we demonstrate a dramatic impact of symmetry: we show that symmetric 2-counter machines are not Turing-complete, by proving decidability of their reachability problem. We conjecture that the decidability extends to machines with 3 or more counters that are symmetric, meaning the set of update transitions is invariant under any permutation of counters. We leave this problem open for machines with 3 and more counters, while identifying a frontier of the impact of symmetry: we prove undecidability for 3-counter machines invariant under the cyclic permutation group of order 3.

0.22

MergeShuffle: Reconstruct Shared Libraries for Improved Performance and Security

cs.SE ·2026-10-16 ·Huaifeng Zhang, Ahmed Ali-Eldin
position
Abstract

Debloating shared libraries has been an active research topic for years. Tools for shared library debloating rely on either static or dynamic analysis techniques to discover which parts of the shared library is used. Once these parts are found, these tools take one of two routes, either removing parts of the library that are not used, requiring reconstruction of the library; or they simply overwriting the unused functions. However, recently a broad study~\cite{brown2024broad} on debloating tools has shown that some of these tools either increase the static binary size, or are generally less effective for complex workloads. In this paper, we argue that shared-library debloaters' performance can be improved by proper reconstruction of the shared-libraries once the used functions in the library are identifies. We propose \myapp, a novel tool for reconstructing shared libraries from a given set of file ranges to eliminate unused code based on user-specified requirements, without requiring access to source code. \myapp takes as input a shared library and a set of file ranges to retain, and performs a two-phase reconstruction: merging nearby ranges to reduce loading overhead, and shuffling them to minimize file size while preserving page-aligned memory mappings. The shuffling problem is formulated as a variant of the Traveling Salesman Problem and solved using a heuristic algorithm. \myapp utilizes program headers to remap retained code to its original memory addresses, preserving user-specified functionality. Our results show that \myapp achieves up to 60\% reduction in file size, 16\% reduction in memory usage, and 32\% reduction in runtime, while maintaining user-specified functionality and improving security.

0.22

Efficient Program Reduction for Library Fuzzing Harnesses

cs.SE ·2026-10-16 ·Haiying Li, Yunlong Lyu, Peng Chen, Hao Chen
code/data
Abstract · 21 pages, 8 figures, 5 tables

Library fuzzing harnesses translate generated inputs into API calls. When a complex harness triggers a crash, developers must determine whether the cause is a library defect or invalid harness usage. A smaller reproducer can simplify this investigation, but obtaining one requires repeatedly building and executing modified harnesses and analyzing their crashes, making manual or even automatic minimization time-consuming. Moreover, data-dependent statements prevent the reducer to remove unnecessary code and the heavy stack trace symbolization in libraries also slows down the candidate execution. To address these challenges, we present HarnessMinimizer, a system for efficient minimization of C/C++ library fuzzing harnesses. Its optimized pipeline reuses header analysis and prepared libraries and runtimes to reduce the work required to build and execute each simplified harness during the reduction process. It records and replays input values so that unnecessary input operations can be removed without changing the values supplied to surviving calls. It also uses a fast method to check whether each simplified harness preserves the original crash, avoiding repeated stack trace symbolization. We evaluate HarnessMinimizer with four reduction engines on 200 crash cases from 21 open source libraries. Compared with the unoptimized pipeline, HarnessMinimizer achieves speedups of 3.46-4.51 times in average total wall time. Average harness size decreases from 1,352 original tokens to 143-263 remaining tokens after minimization. In crash triage using a large language model, accuracy in identifying harness bugs increases from 28% with the original harnesses to 44-51% with the minimized harnesses, while accuracy for library bugs remains at 100%.

0.20

Who Verifies the Builder? Toward Independent Assurance for Agent-Produced Software

cs.SE cs.AI ·2026-10-16 ·Sion Israel Sion, Alain April
code/data
Abstract · 10 pages, 2 figures, 4 tables. Artifact: https://github.com/besionisrael/assure-harness

Coding agents can now implement a specification, write and run tests, repair failures, and declare the result ready for release. When the same agentic system produces both the software and the evidence of its correctness, a passing test suite is no longer independent evidence: visible tests become an optimization target, code and tests may share one misinterpretation of the specification, and the builder's own tests may simply be incomplete. This paper argues that agent-produced software requires an assurance plane that is structurally independent of the build plane. We characterize four gaps between stakeholder intent and production (requirement, conformance, evaluation, and model gaps) and identify which of them verification can close. We propose a framework that separates building from verifying, introduces a Verification Contract stating what must be demonstrated before release, and organizes verification in a specification-first phase without access to the implementation and an implementation-aware phase based on mutation analysis, property-based testing, and fuzzing, with an assurance report that leaves the release decision to a human. We evaluate the framework in a pre-registered study on three stateful services, two of them written by third-party models, and an EvalPlus pilot. The builder's programs were correct, but its own tests missed subtle interactions: a verifier in a fresh context detected more of the 46 seeded faults than the builder's tests (Holm-corrected p = 0.006) and surfaced the ambiguities of the specification, at the price of rare false oracles. A verifier from another provider did not add detection.

0.20

An Interactive GraphRAG Workspace for Reifying, Auditing, and Steering Conversational AI Memories

cs.SE ·2026-10-16 ·Munshi Md Arafat Hussain, Md. Abdur Rahman, Md Saeed Siddik
Abstract

Conversational Large Language Models (LLMs) are widely employed in software engineering for code generation, debugging, and architectural design. However, as developers engage in multi-session development tasks, technical knowledge becomes siloed in isolated, linear chat histories. Reusing context across sessions leads to three critical challenges: (1) flat-vector retrieval fails to traverse multi-hop relational dependencies between distant architectural decisions; (2) temporal drift in software packages and APIs results in silent context poisoning when reusing aged dialogue logs; and (3) raw conversation concatenation induces prompt bloat and attention degradation. We present CognitiveCanvas, an interactive tool that transforms conversational AI memories from ephemeral text logs into persistent, manipulable, and audited knowledge objects. \textsc{CognitiveCanvas} integrates: (1) a Hierarchical Semantic Fact-Graph (HSFG) that distills dialogues into Goal, Fact, and Code artifacts, achieving a 92.2\% token reduction while enabling multi-hop GraphRAG traversal; (2) a Continuous Temporal Drift Auditor coupling 60+ regex heuristics with live calendar-anchored LLM verification for 1-click self-correction; and (3) a Spatial-Symbolic Prompt Compiler (SSPC) that translates 2D direct manipulation gestures (Voronoi clustering, node resizing, synthesis edges) directly into structured attention directives

0.18

VulnSAGE: Adjudicating SAST Reports by Connecting LLM Reasoning to Program Structure

cs.SE cs.CRcs.PL ·2026-10-16 ·Jinbao Chen, Zixiang Zhou, Yu Zhang, Yihong Wu
Abstract

Static Application Security Testing (SAST) reports flag operations whose safety depends on project-specific context, and maintainers must decide which reports describe real defects. Large language models (LLMs) can interpret this context, but adjudicating a report with them requires the decisive code, which often lies far from the reported line, and a decision basis that program analysis can check. We present VulnSAGE, an adjudication system for SAST reports that connects LLM reasoning to program structure in both directions. Its report-centered slicer recovers the dependencies of the reported operation together with the program structure that explains them, such as the caller object on which a helper acts. CLAIM, an intermediate representation for report judgments, records each model interpretation as a semantic fact that cites the program relations it assumes. Program analysis checks these premises, and the verdict follows from the facts whose premises hold. Both mechanisms share a program graph, which lets the model cite the code it reads and lets the checker query those citations. Statically supported true positives then guide proof-of-concept generation and runtime reproduction. Across 187 labeled reports from four corpora covering five vulnerability families, VulnSAGE rejects 89 of 97 false alarms and retains 85 of 90 real defects. Under a common PoC procedure, 25 of its positive verdicts reproduce at runtime, compared with three for the LLM-based adjudicator LLM4SA. Applied to 45,440 alerts raised by five SAST tools on the 200 most-starred C/C++ projects on GitHub, it reproduces 104 defects at runtime. Maintainers have confirmed 24 of the 26 defects they have responded to, including ten previously unknown defects, nine of which are already fixed.

0.18

Recall-Oriented Precise Binary-to-Source Function Matching

cs.SE ·2026-10-16 ·Junlin Li, Jianjun Huang, Wei You, Bin Liang
Abstract · 21 pages, 5 figures

Binary-to-source function matching supports software composition analysis by identifying reused third-party code, but existing embedding-based approaches restrict subsequent refinement to initially retrieved candidates. When the correct source function is absent from this set, selecting or filtering the available candidates cannot recover the match. Inlining further complicates the retrieval by combining multiple source functions into one binary function. In this paper, we propose BScout to improve matching precision and uncover additional matches. BScout uses the positional information of existing matches to supplement candidates for unresolved binary functions, extending the search beyond the initial candidate set, and adopts composition-aware training to address the inlined composition matching. Large language models are adopted to select and review matches, while positional information also helps filter inconsistent results. Experiments show that BScout is effective in binary-to-source function matching. BScout achieves overall higher precision, recall and F1 scores than BinaryAI and SBridge with acceptable analysis costs. Particularly, on large stripped binaries, BScout outperforms the competitors with a precision increment of 32% and approximately 2.9 times the recall.

0.18

Towards Agentic Proof Harness Engineering

cs.SE ·2026-10-16 ·Rohith Kumar Shanmuganathan, Cedric Richter, Heike Wehrheim
Abstract · NIER paper, under submission, 16 pages

Formal verification of real-world projects requires substantial manual effort, from specifying correctness properties to constructing the infrastructure needed to check them. We envision verification agents that assist developers throughout this process. Among the many tasks involved, one stands out as both foundational and amenable to automation: proof-harness engineering. A proof harness defines an isolated verification environment for a program unit, such as a function or module, by providing symbolic inputs, initialization code, and models of its dependencies. Today, proof-harness engineering remains largely a manual task performed by verification engineers. To explore the potential of coding agents for this task, we propose an agentic proof-harness engineer that automatically constructs verifier-ready harnesses from a software project and a designated verification target. We combine the agent with formal soundness and precision guards to validate generated abstractions and guide their refinement. We evaluate the feasibility of our approach in an exploratory case study on curl and libuv, using three historical assertion violations already embedded in the source code as verification properties. Our results show that the agent successfully constructs proof harnesses that enable CBMC to detect all three violations. We further demonstrate the potential for reusing these harnesses with other verification tools.

0.16

A generalised and modular proof-theoretic semantics for sequent systems

cs.LO math.LO ·2026-10-16 ·Victor Barroso-Nascimento, Arghya Chakrabarty, Elaine Pimentel
Abstract

Base-extension semantics explains logical consequence via derivability in atomic systems. Extending this approach across logics requires careful control of structural principles and contexts. We present a modular sequent-based framework in which support is indexed by multisets of atomic resources and the semantic clauses for connectives are guided by their left introduction rules. The framework separates additive from multiplicative behaviour and makes explicit the contributions of structural rules and restrictions on sequents. We establish soundness and completeness for classical and intuitionistic propositional logic and for multiplicative-additive fragments of classical and intuitionistic linear logic. We then investigate axiomatic extensions of intuitionistic logic through Jankov logic. Two classes of atomic bases yield sound and complete semantics. Completeness is proven using two novel and modular methods using simulation bases, one of which is constructive. These results provide a common semantic treatment of several logical systems while identifying the additional conditions needed to extend the framework beyond structural variations.

0.16

Arbitrary Functors as Modalities in a Linear Calculus

cs.PL ·2026-10-16 ·Jean-Philippe Bernardy, Andreas Abel
Abstract

We study modalities in the linear lambda calculus, by interpreting them as endofunctors in monoidal closed categories. We support functors that do not distribute over products, including (co)monads. We also support arbitrary natural transformations as conversion between modalities. We provide a generic categorical semantics, which we then specialise to a sound and complete resource semantics. We use this resource semantics to derive normalisation by evaluation. Along the way, we describe several applications of such models. This work aims at providing a foundation for future programming languages with modalities. The paper is backed by a mechanisation in Agda.

0.15

Enforcing Code Reasoning in Agentic Scaffold for More Effective Issue Resolution

cs.SE ·2026-10-16 ·Changshu Liu, Ji young Kim, Alireza Ghazanfari, Saanvi Nallu, Reyhaneh Jabbarvand
Abstract

Reasoning is central to ReAct-based programming agents, yet whether and how they reason about code remains unclear. We study 7,955 issue-resolution trajectories from six single- and multi-agent scaffolds across four programming languages on SWE-bench-Verified, SWE-bench-Pro, and SWE-bench-Lite. We identify 12 forms of code reasoning in 66.95% of trajectories. Code reasoning is associated with higher resolution rates, and its intensity increases with task complexity. However, it remains largely unguided: agents may focus on surface-level symptoms during localization, fail to refine near-correct patches, overlook issue requirements, and generate tests with limited behavioral diversity. To address these limitations, we propose DeskCheck, an agentic scaffold that systematically guides code reasoning throughout repository-level issue resolution. DeskCheck simulates execution paths to localize root causes, simulates candidate patches to refine near-misses, audits patches against issue requirements, and reasons about corner-case inputs and expected outputs before generating tests. On 731 SWE-bench-Pro instances, DeskCheck improves resolution rates over mini-SWE-Agent and Claude Code by 6.29% to 10.53%, at an additional cost of $0.02-$0.28 per instance (average: $0.14). Per-phase analyses show higher method-level localization recall, more successful repairs of imprecisely localized issues, and better detection of missed corner cases in near-correct patches. Exact McNemar tests show statistically significant improvements across all subject LLMs (p < 10^-10), while repeated experiments support robustness to agent nondeterminism. These findings motivate systematic code reasoning as a means of improving programming agents.

0.14

Empirical Insights from AI-Powered Software Engineering in Practice

cs.SE ·2026-10-16 ·Tina Mersch, Jack Johns, Steffen Klepke, Hanya Elhashemy, Artur Signell +2 more
Abstract · Submitted to IEEE Software

Generative Artificial Intelligence (GenAI) is becoming indispensable in future software engineering. Its application spans a wide spectrum - from assisted development to autonomous software engineering, requiring organizations to address associated risks and challenges. This paper investigates these issues through three case studies, highlighting practical problems encountered during GenAI integration. Based on these findings, it provides targeted recommendations for managing risks and ensuring reliable outcomes. This paper provides guidance for effective and responsible GenAI adoption in software engineering based on empirical insights.

0.11

A Taxonomy of Predicate Transformer Monads for Specifying Monadic Programs

cs.PL ·2026-10-16 ·Abigail Pribisova, Yao Li, Cezar-Constantin Andrici, Catalin Hritcu, Kenji Maillard +1 more
Abstract · Submitted to ESOP'27, in v2 same PDF but fixed author order in metadata

Monads of weakest precondition predicate transformers are the foundation of several verification frameworks for monadic programs. For specifying monadic computations with nondeterminism and nontermination, in addition to weakest preconditions, we introduce the first strongest postcondition monads covering all combinations of angelic or demonic nondeterminism and partial or total interpretations of nontermination. For extending these base strongest postcondition monads with side effects such as mutable state, monad transformers do not yield the expected monads, so we instead generalize the base monads to arrows and use arrow transformers. We give a taxonomy of weakest precondition and strongest postcondition monads, for the base monads and for their versions with state, and show that the known relationships between under- and over-approximate triples based on predicate transformers already hold at the level of the predicate transformers themselves. We use the predicate transformer monads to define under- and over-approximate triples and, where possible, Dijkstra monads. Finally, we relate these two specification styles.

0.06

Integer Reachability in Data VASS

cs.FL ·2026-10-16 ·Roland Guttenberg, Łukasz Kamiński, Sławomir Lasota
Abstract

Data Petri nets are an extension of plain Petri nets where, intuitively speaking, tokens carry data values, and transitions may perform equality tests on data values involved. Likewise, data vector addition systems with states (data VASS) are an extension of plain VASS, and equivalent to data Petri nets. The decidability status of the reachability problem in this model is an established and long-standing open question. In this paper we perform a major step towards resolving this question: we show decidability of the integer reachability problem, where the counters can drop below zero. Contrarily to plain VASS, where integer reachability reduces to integer linear programming, the setting of data VASS is significantly more complex, mostly due to the fact that the state spaces are not finite but orbit-finite (like in register automata). Our decision procedure combines an iterative refinement, resembling the KLM decomposition in plain VASS, with a termination argument via a novel way of bounding the number of data values appearing in the refinement. As corollaries, we deduce decidability of a number of decision problems concerning register automata and data Petri nets.

0.06

Theory & Algorithms 64

Revisiting separators in string graphs, region intersection graphs, and induced-minor-free graphs

cs.CG math.COcs.DMmath.MG ·2026-10-16 ·James Davies, Paweł Rzążewski, Jack Spalding-Jamieson, Bartosz Walczak
top venuetheoryhighly cited
Abstract · To appear at SODA 2027

String graphs are intersection graphs of continuous curves in the plane. In 2010, Fox and Pach proved that string graphs with $m$ edges admit balanced vertex separators of size $\mathcal{O}(m^{3/4}\sqrt{\log m})$ and conjectured that this bound can be improved to $\mathcal{O}(\sqrt{m})$, which would be optimal. After partial progress, Lee claimed in 2017 to prove this conjecture, and more generally to prove analogous separator theorems for region intersection graphs over minor-closed graph families. A key intermediate step in Lee's approach was a padded decomposition scheme for region intersection graphs. In 2024, Korhonen and Lokshtanov extended Lee's approach to the broader setting of graph classes excluding a fixed graph as an induced minor. These results have since found applications in a wide range of works. Later in 2024, a flaw was discovered in the construction of Lee's padded decomposition scheme, invalidating Lee's as well as Korhonen and Lokshtanov's separator results, and many works relying on them. In this work, we prove that string graphs with $m$ edges admit balanced separators of size $\mathcal{O}(\sqrt{m})$, and more generally obtain analogous $\mathcal{O}(\sqrt{m})$ bounds for region intersection graphs over fixed minor-closed graph families and for graphs excluding a fixed graph as an induced minor, thereby repairing the vast majority of the affected works. Furthermore, our balanced separators can be found in polynomial time. Our proof uses a different approach from Lee's and does not rely on padded decompositions. Since padded decompositions have also found applications independent of the separator theorems, particularly for string graphs, we also prove a padded decomposition result for region intersection graphs that is optimal for string graphs. We have formalized both the balanced separator and the padded decomposition results in Lean 4 with mathlib.

0.70

Competitive Online Clustering with Movement Cost

cs.DS ·2026-10-16 ·Christian Coester, Yichen Huang, Michael Mitzenmacher
top venue
Abstract · To appear in SODA 2027

We study the problem of online clustering with movement cost, which generalizes the fundamental $k$-server problem. At each time step, the algorithm is given a set of points in a metric space and must choose $k$ cluster centers. The cost incurred is the sum of the clustering cost at each time step and the total distance that the centers move over time. Prior work established competitive algorithms for the case $k\in\{1,2\}$ on the line metric with the $k$-median clustering objective; under the $k$-median objective, this problem is also known as facility reallocation. We substantially generalize these results to arbitrary $k$, general metric spaces, and a broad class of clustering objectives. Specifically, we consider clustering costs defined as a norm of the distance vector from points to their nearest center, which subsumes the $k$-median, $k$-center, and (square root of) $k$-means objectives, and provide an asymptotically optimal $O(k)$-competitive randomized algorithm against adaptive online adversaries. The algorithm is based on a reduction to the $k$-server problem, establishing a tight connection between the problems. In this reduction, $k$-server requests are generated by Poisson processes, whose arrival rates evolve in continuous time and are driven by a dual representation of the clustering costs. Our method of using Poissonization in this context may also be of independent interest. We demonstrate another use of the idea by deriving similar reductions for batched set chasing problems. We also consider a different family of clustering objectives defined as a norm of the vector of cluster radii, which subsumes the sum-$k$-radii and the $k$-center objectives, and give a tight $O(k)$-competitive deterministic algorithm on the line.

0.62

Anytime Contracts for AI Services

cs.GT cs.AIcs.LG ·2026-10-16 ·Boris Velasevic, Eden Saig, Eric Mazumdar
Abstract · 19 pages, 3 figures

AI services are prone to moral hazard, as model providers can silently route requests to cheaper, lower-quality models, and the inherent stochasticity of model outputs makes this degradation hard to detect. Contract theory addresses moral hazard by designing payment rules that reward high-quality outputs and prevent strategic overcharging. However, existing frameworks assume interactions of fixed duration, whereas AI services are typically consumed through subscriptions and pay-as-you-go APIs that users can terminate at any time based on observed performance. We show that this flexibility can break standard contracts, opening them up to abuse by either the providers or users. To address this, we extend contract theory to account for unknown exit rules. We formalize the problem as a Principal-Agent game, and show that budget-optimal contracts in this setting are proportional to anytime-valid hypothesis tests based on likelihood-ratio e-values. We give an efficient algorithm for computing anytime contracts and show how betting e-processes can be used to approximate them when the models are unknown. Finally, we evaluate our findings on synthetic data and data from deployed LLMs.

0.58

Extending Myerson's Optimal Auctions to Correlated Bidders via Neural Network Interpolation

cs.GT cs.LGecon.TH ·2026-10-16 ·Mingyu Guo, Jiayuan Liu, Vincent Conitzer
top venuetheory
Abstract · Accepted at NeurIPS 2026

We aim to design revenue-maximizing single-item auctions that are deterministic, strategy-proof, and ex post individually rational --- the quintessential fundamental model for optimal mechanism design. While Myerson's seminal work solved this model for independent bidders, straightforward extensions to correlated settings often result in non-monotonic, and therefore non-strategy-proof allocations. Critically, optimal allocation design for correlated bidders is NP-hard to approximate, rendering theoretical optimality guarantees in this domain computationally intractable. We establish a new standard of empirical rigor for this theoretically hard model by proposing an empirical pipeline. We synthesize neural interpolation of Myerson's greedy allocation based on marginal profits (the correlated variant of virtual valuation), formal neural verification to enforce exact strategy-proofness, and a constructive repair procedure. Empirically, our method is highly effective, consistently achieves near-optimal revenue across a wide range of distributions, including synthetic, adversarial, and real-world (consumer, financial, and industrial) datasets. Compared to existing manual and neural baselines, our approach shows substantial improvement, often reducing the revenue gap by an order of magnitude. Our study is also the first to introduce (unattainable) greedy revenue as a rigorous upper bound for empirical benchmarking, providing a definitive quantitative performance measure where true optima remain theoretically elusive. We demonstrate the generality of our approach by extending it to multi-unit auctions with unit demand and integrating our verification techniques into RegretNet to achieve exact strategy-proofness.

0.56

Representing Information on DNA using Patterns Induced by Enzymatic Labeling

cs.IT ·2026-10-16 ·Daniella Bar-Lev, Eitan Yaakobi, Zohar Yakhini, Tuvi Etzion
published
Abstract

Enzymatic DNA labeling is a powerful tool with applications in biochemistry, molecular biology, biotechnology, medical science, and genomic research. This paper contributes to the evolving field of DNA-based data storage by presenting a formal framework for modeling DNA labeling in strings, specifically tailored for data storage purposes. Our approach involves a known DNA molecule as a template for labeling, employing patterns induced by a set of designed labels to represent information. One hypothetical implementation can use CRISPR-Cas9 and gRNA reagents for labeling. Various aspects of the general labeling channel, including fixed-length labels, are explored, and upper bounds on the maximum size of the corresponding codes are given. The study includes the development of an efficient encoder-decoder pair that is proven to be optimal in terms of maximum code size under specific conditions. The codewords for this framework are codesets, i.e., each codeword is a collection of a few labels. Of special interest are codes whose codesets contain fixed-length labels. Their construction and analysis are tied to de Bruijn sequences and constrained coding.

0.55

UAV Rotatable Antenna under Rotation Delay

cs.IT eess.SP ·2026-10-16 ·Yinmei Xie, Chiya Zhang, Chunlong He, Weizheng Zhang, Wei Wang
Abstract · 13 pages, 10 figures, 1 table

Rotatable antennas (RAs) can improve wireless link performance by adjusting their boresight directions. However, existing RA studies generally assume instantaneous antenna reconfiguration, whereas mechanical rotation incurs non-negligible delay in practical deployment. To address this issue, we develop a rotation-delay-aware communication model for finite communication slots, where mechanical rotation directly reduces the time available for data transmission. This model reveals a fundamental tradeoff between the communication gain achieved through RA reorientation and the associated transmission-time loss. Accordingly, we formulate a rotation-delay-aware joint design of RA orientations and transmit beamforming to maximize the minimum effective transmission rate. For the single-user system, under a sufficient condition for strict concavity, the global optimum falls into one of three regimes: no rotation, partial rotation, and full rotation. A closed-form approximation is derived using a second-order expansion. For the multi-user system, we separate rotation-budget selection from fixed-budget max-min signal-to-interference-plus-noise ratio (SINR) optimization. An analytic-gradient zero-forcing (AG-ZF) algorithm is then developed for low-complexity joint RA orientation and rotation-budget optimization. The results show that explicitly accounting for mechanical rotation delay is essential for finite-slot RA-enabled UAV communications, and that the proposed designs effectively balance reconfiguration gain and transmission-time loss while achieving favorable effective-rate performance with low computational complexity.

0.51

Unified Geometric--Kinematic Laws of Mobility in 3D Cellular Networks

cs.IT ·2026-10-16 ·Tian Shi, Minghua Xia
top venue
Abstract · 15 pages, 8 figures, 2 tables, to appear in IEEE TWC

Mobility in three-dimensional (3D) cellular networks induces frequent handovers and random sojourn times for aerial users. Analyzing these events is challenging because they jointly depend on user trajectories and random service-boundary geometry. This paper develops a unified geometric--kinematic framework in which service regions form a stationary random tessellation, their boundaries constitute a random closed set, and user motion follows a cruise--turn mobility model. Under the stated stationarity, isotropy, and independence conditions, the handover rate satisfies the exact law \(H=(S_V/2)\bar v\), where \(S_V\) is the specific surface area and \(\bar v\) is the long-run mean speed. The forward boundary-distance distribution provides the local geometric input to an effective-traversal approximation for the mean remaining sojourn time. These results link global and local geometric descriptors to complementary mobility metrics. Combining the handover law with link coverage yields a handover-aware effective spectral efficiency for base station (BS) density selection and service-architecture comparison. We then specialize the framework to Poisson--Voronoi single-BS service and Poisson--Delaunay four-BS coordinated multi-point service; under both models, the handover rate scales as \(λ^{1/3}\bar v\). Numerical results corroborate the exact handover expressions, assess the sojourn-time approximations, and show how user speed and interruption duration affect the preferred density and architecture.

0.49

Integrated sensing, communications, and identification for low-altitude economy

cs.IT ·2026-10-16 ·Shanxing Zeng, Jun Wang, Ying-Chang Liang
published
Abstract · 37 pages, 5 figures, 4 tables. Accepted for publication in SCIENCE CHINA Information Sciences

With the rapid development of low-altitude economy (LAE), low-altitude information networks (LAINs) must provide reliable communication, environmental sensing, and target identification to support efficient and safe airspace operation. Although integrated sensing and communications (ISAC) provides a foundation for the joint design of communication and sensing, identification must also be incorporated to meet airspace safety and regulatory requirements. This paper reviews integrated sensing, communications, and identification (ISCAI) for LAE. We first review low-altitude communication, sensing, and identification approaches and explain how representative applications motivate their integration. We then summarize low-altitude ISAC fundamentals, including system architectures, channel modeling, unified signal models, and performance metrics, followed by communication-centric and sensing-centric designs, resource allocation, and aircraft trajectory optimization. Building on these, we introduce ISCAI, focusing on a backscatter communication (BackCom)-based identification mechanism in which backscatter devices serve as electronic license plates for aircraft. We present system and signal models for integrated sensing and identification (ISAI) and ISCAI, together with multifunctional receiver designs and mutual integration gains. Finally, we discuss open problems and future directions, including an information-theoretic framework, artificial intelligence-empowered system design, and reconfigurable intelligent surface-enhanced ISCAI. This review provides a structured perspective on the evolution of LAINs toward integrated communication, sensing, and identification.

0.46

Online geometric matching and segments block points

cs.DS cs.CG ·2026-10-16 ·Yaqiao Li, Junyi Lin
publishedhot direction
Abstract · accepted by IJTCS-FAW 2026

Given a sequence of online points in general position in the plane, the geometric matching problem asks to match as many points as possible, subject to segment crossing constraints. We initiate the study of online geometric matching with limited crosses. In the $k$-crossing model each matched segment is allowed to have at most $k$ crosses, where $k=0$ corresponds to the online noncrossing matching model for which Bose et al. proved that Greedy is an optimal deterministic algorithm achieving a competitive ratio of $2/3$. In the triangle-free-crossing model no three matched segments are allowed to pairwise cross each other. Interestingly, we show in both cases the performance of Greedy does not improve by allowing crosses, and in fact, becomes worse in the $k$-crossing setting for $k\ge 1$. In the triangle-free-crossing model, we propose a simple two-layer-Greedy deterministic algorithm achieving a competitive ratio $4/5$. More interestingly, we propose and study the underlying segments blocking points extremal problem, and discuss its connection to vertex cover numbers and Ramsey numbers, with interesting open problems that might be of independent interest. We revisit the noncrossing model. We give a randomized algorithm which is an asymmetric version of Kamali et al., improving the competitive ratio from $0.66952$ to $0.67034$. We also show, via a conceptually simple construction, that no randomized algorithm can match more than $6/7$ fraction of points, improving the previous $16/17$ by Boyar et al.

0.40

On the Length of Maximum Sum-Rank Distance Codes

cs.IT math.CO ·2026-10-16 ·Valentina Astore, Ferdinando Zullo
Abstract

The classical theory of arcs in finite projective spaces provides a powerful geometric framework for studying maximum distance separable (MDS) codes and their possible lengths. In this paper, we investigate how this perspective extends to maximum sum-rank distance (MSRD) codes, whose geometric description involves systems of linear sets rather than individual projective points. Our main goal is to understand how the block structure of an MSRD code constrains its length. In dimension three, we determine the possible intersection patterns of lines with the associated linear sets and obtain explicit formulas for their characters and the weight distribution of the corresponding codes. These results lead to new length bounds and show that the block structure comes at a cost: the presence of a block of rank at least two reduces the maximum possible total length by at least $q-1$, and sufficiently long MSRD codes must be classical, that is, MDS codes over $\mathbb F_{q^m}$. We also prove that three-dimensional one-weight MSRD codes do not exist. In arbitrary dimension, we interpret shortening geometrically as projection and establish a general projection bound that extends classical counting arguments for arcs. This yields restrictions depending explicitly on the block profile and, for full-length blocks, identifies parameter ranges in which linearized Reed--Solomon codes attain the maximum possible number of blocks. Finally, we construct an exceptional three-dimensional MSRD code over $\mathbb F_9$ with three blocks of length two and propose a sum-rank analogue of the MDS conjecture.

0.39

Orientation Diversity Provably Increases the Angular-Domain Rank of 3D MIMO Arrays

cs.IT ·2026-10-16 ·Samer Henry, Elvino Sousa
publishedtheory
Abstract · 5 pages, 4 figures. Accepted for publication in IEEE Wireless Communications Letters

Movable and rotatable-antenna systems treat element placement as a design resource, but the supporting theory is largely position-only. We prove that element orientation is an effective spatial resource in its own right, complementary to position. For collinear or coplanar arrays, the angular-domain rank is capped by the number of projection-ambiguity classes of the scattering directions, for every common element orientation and pattern. Orientation-diverse elements with sufficiently rich analytic patterns restore the maximal rank min(N,L) for almost every orientation draw. A closed-form correlation expression, strictly monotone in orientation spread under isotropic scattering, quantifies the gain within the scalar-pattern model, and Monte Carlo simulations corroborate the exact-rank and capacity behavior.

0.39

Batch range mode in O(n+Q)^1.395

cs.DS ·2026-10-16 ·Filippo di Gravina
published
Abstract · 18 pages, 0 figures

We give a deterministic worst-case algorithm for exact Batched Range Mode on an arbitrary array of length n and a batch of Q intervals. The algorithm returns the mode frequency and the smallest mode identifier for each query in time (n+Q)^(tau+o(1)), where tau = 5 - sqrt(13) = 1.394448724536... . The algorithm combines virtual occurrence windows and block-boundary snapping with a query-restricted min-plus product whose first operand is column-monotone and whose second operand is row-monotone. A deterministic modular verifier uses geometric and query potentials to bound the same active regions; false modular matches are corrected by offline orthogonal range updates. The final analysis uses the sharper rectangular exponent curve of Uznanski (2026), together with exact rational parameter balancing and a transfer to prime fields. The analysis covers every frequency scale and arbitrary batch sizes. Exact frequencies and the smallest mode identifiers are recovered without increasing the exponent.

0.38

Counting and Sampling Manufacturable Stacking Sequences

cs.DM cs.CEcs.FL ·2026-10-16 ·Alessio Langiu, Felice Rubino, Giacomo Canale
Abstract · 22 pages, 3 figures, 4 tables. Code: https://doi.org/10.5281/zenodo.23391635

Designing the stacking sequence of a composite laminate is a discrete optimisation problem: each ply takes one of a few orientations, each evaluation may cost a finite-element solve, and the design must obey manufacturing rules and design guidelines, which optimisers enforce by penalty, repair or rejection. We describe the admissible set exactly. The two manufacturing rules forbid a finite set of factors, so the sequences they admit form a regular language, recognised by an automaton that tracks the last orientation and the length of its run; balance and the ten per cent rule constrain the Parikh vector and are carried by a dynamic program over automaton state and counters. One table, polynomial in N, counts the admissible sequences without producing them, draws them uniformly, and tells whether a partial sequence can still be completed. We give pseudocode, exact counts for three orientation sets, closed-form ceilings on the composition, and four of the five single-rule entries a published table left uncomputed. At 48 plies, 99.8% of the nodes of a layer-wise search tree checking the rules locally have no admissible completion on eight orientations; the completion test removes them. A one-point crossover that cuts two admissible parents only where they share an automaton state and their child closes the counters returns admissible children by construction.

0.37

Spatial Distribution of Mean PAoI in LEO Satellite-enabled IoT Networks

cs.IT ·2026-10-16 ·Badiaa Gabr, Mustafa A. Kishk
top venue
Abstract

Non-terrestrial networks (NTNs), such as Low Earth Orbit (LEO) satellites, can provide consistent coverage for Internet of Things (IoT) nodes, particularly in rural areas where cellular coverage is lacking. These IoT nodes are used to monitor and record their surroundings, supporting various applications, including real-time applications. In such cases, transmission of the most up-to-date data is essential, making the Age of Information (AoI) a critical metric to consider. The AoI measures the elapsed time from the data generation at the source node to its reception at the destination. Moreover, the Peak AoI (PAoI), which denotes the maximum AoI in the system before an update is received, provides valuable insights into the network's performance. Therefore, this encourages us to focus on the PAoI within a large-scale and fully loaded IoT network, which is covered by LEOSATs. Using tools from stochastic geometry, this paper derives the spatial distribution of the uplink mean PAoI in a LEOSAT-enabled IoT network. To achieve that purpose, the meta distribution of the uplink conditional coverage probability is derived. Numerical results articulate how varying the number of deployed LEOSATs and their altitudes affects the coverage probability meta distribution and the spatial distribution of the mean PAoI. Additionally, by examining various scenarios with different numbers of LEOSATs and altitudes, we can derive valuable insights to determine the optimal beamwidth for LEOSATs tailored to each specific scenario to minimize the conditional mean PAoI.

0.37

Optimally Curating an Event

cs.GT ·2026-10-16 ·MohammadTaghi Hajiaghayi, Sebastien Lahaie, Mohammad Mahdavi, Suho Shin
top venue
Abstract · Appeared at IJCAI 2026

We study the optimal design of a self-financing event, a problem that requires balancing the recruitment of costly, positive-value participants with revenue-generating agents who may impose negative values on the event. We introduce a novel two-sided mechanism design framework with plus agents equipped with private costs and positive impact, and minus agents with private values and negative impact upon inclusion in the event, to maximize the overall quality of the event under a budget-balanced (BB) constraint so that the designer does not run a deficit. We conduct a comprehensive study on the theory of optimal event curation on various utility functions. For additive utility, we fully characterize the optimal incentive-compatible Bayesian mechanism under both the ex-ante and ex-post BB constraints. For submodular utility, we propose an ex-ante BB mechanism that achieves a constant-factor approximation to the optimal Bayesian mechanism.

0.36

Uniform-Price Mechanism Design for Automated Market Makers

cs.GT ·2026-10-16 ·Zhike Peng
published
Abstract · 22 pages. Full version of a paper accepted at IJTCS-FAW 2026

Automated market makers (AMMs) are central to decentralized exchanges, but their reserve curves make mechanism design different from classical double auctions. We study a batch double auction with buyers, sellers, and a liquidity-nondecreasing AMM: post-trade reserves must have weakly larger product than the initial reserves. Each user has two-dimensional private information: a unit value and a demand or supply quantity. User trades clear at one uniform price subject to AMM feasibility. The objective is user-side trading volume, benchmarked against the optimal feasible individually rational (IR) uniform-price allocation. We give a randomized uniform-price mechanism based on binary quantity buckets. It randomly chooses one of three clearing modes: peer-to-peer matching, AMM sales to buyers, and seller sales to the AMM. Every deterministic realization of the mechanism is ex-post feasible and satisfies IR and weak quantity incentive compatibility: conditional on any quantity report, truthful unit-value reporting is dominant whenever it gives a non-negative utility, no unit-value deviation gives a positive utility if the truthful one doesn't, and users cannot benefit by under-reporting the quantity together with their truthful unit value. Let $V^\ast$ be the optimal feasible IR uniform-price volume without incentive constraints, and let $γ=\bar q/\underline q$, where $\bar q$ and $\underline q$ are the quantity-domain bounds. For $γ\ge1$, the mechanism achieves an expected trading volume of $ Ω(1/(1+\logγ)^2)V^\ast - O(\bar q). $ We also show that full incentive compatibility over the two-dimensional type space forces a worst-case approximation ratio of $O(1/γ)$ at every prescribed initial reserve ratio, even when both AMM reserves are scaled by an arbitrarily large common factor and the optimal volume $V^\ast$ is arbitrarily large relative to $\bar q$.

0.36

Proportionality in Ranking Compression

cs.GT cs.LGecon.TH ·2026-10-16 ·Daniel Halpern, Xinyu Liu, Evi Micha, Yu Peng Ng, Warut Suksompong
top venuetheory
Abstract · Appears in the 40th Annual Conference on Neural Information Processing Systems (NeurIPS), 2026

Motivated by applications in reinforcement learning from human feedback, we consider a setting where a given collection of rankings needs to be compressed into a smaller collection of rankings. This can also be seen as a combination of two fundamental social choice frameworks: ranking aggregation and multiwinner voting. We propose three proportionality notions that capture different types of representation in this setting: positional proportionality, pairwise proportionality, and proportionality for solid coalitions (PSC). On the one hand, we show that positional proportionality and PSC are always satisfiable for any input rankings and target compression size, and a desired output can be found in polynomial time. On the other hand, we prove that pairwise proportionality cannot be satisfied in general, but can nevertheless be attained when the input rankings are single-peaked or single-crossing.

0.33

Polytope Sampling at the Speed of Isoperimetry

cs.DS math.OCmath.PR ·2026-10-16 ·Anming Gu, Bobby Shi, Kevin Tian, Hubert Yang
Abstract · 58 pages

We give faster algorithms for sampling from the uniform distribution over a polytope. Our main result is an algorithm that, when applied to a polytope with $m$ constraints in $n$ dimensions, uses $m \cdot \mathrm{poly}\log(\frac m δ)$ iterations, each running in time $O(mn^{ω- 1})$, to return a sample within total variation distance $δ$ to the uniform distribution. When $m =O(n)$ and $δ= \mathrm{poly}(\frac 1 n)$, our runtime improves upon all prior works by an $Ω(n^{2/3})$ factor. Our sampler is based on the recent \emph{functional stochastic localization} framework of [GST26], and achieves an iteration count nearly matching the Poincaré constant of the uniform distribution with respect to the logarithmic barrier.

0.33

Simpler Algorithms for Capacitated $k$-Median with $(1 + ε)$-Violations Using New Inequalities

cs.DS ·2026-10-16 ·Han Dai, Shi Li
published
Abstract · Accepted at SOSA 2027

We revisit the capacitated $k$-median problem, focusing on constant-factor approximation algorithms with either $(1+ε)$-cardinality or capacity violations. Previous results [DemirciL16, Li16} achieve an $O_ε(1)$-approximation with $(1+ε)$-violations via complex extended configuration linear programming formulations with the round-or-cut framework. We propose a simpler, unified round-or-cut framework that handles both cardinality and capacity violation settings. Our main technical contribution is a mechanism that directly utilizes the costs of local instances to generate valid inequalities. Specifically, we partition the metric space into local pieces of size $O(1/ε)$ as in previous work, and construct instances of capacitated clustering with penalties, which allow client demands to be unserved at a penalty cost. Solving these local instances yields lower bounds on the original LP cost, yielding valid LP constraints. In contrast, previous approaches rely on complex extended configuration LPs over connection variables $x_{ij}$'s and opening variables $y_i$'s. As a secondary contribution, our approach achieves an improved running time of $\text{poly}(n, 1/ε)$, bypassing the exponential dependence on $1/ε$ required to solve the previous extended configuration LPs. Instead of solving the local instances optimally in time exponential in $1/ε$, we obtain $O(1/ε)$-approximate solutions in $\text{poly}(n, 1/ε)$ time. Finally, for the cardinality-violation setting, our algorithm opens a set of at most $(1+ε)k$ $\textit{distinct}$ facilities, and thus gives a bicriteria approximation in the $\textit{hard-capacitated}$ setting. In contrast, the algorithm of [Li16] may open the same facility multiple times.

0.33

Puzznic is PSPACE-Complete

cs.CC cs.CG ·2026-10-16 ·Jake Anderson, Lars Magnus Hvattum, András Z. Salamon
theory
Abstract · 15 pages

We prove that deciding solvability of Puzznic, a falling-block puzzle game originally developed by Taito, is PSPACE-complete. To show hardness we construct a self-closing door gadget using Puzznic's game mechanics. For membership, we show that the configuration space is exponentially bounded and apply Savitch's theorem. We also improve an existing NP-completeness proof for a simpler version of the game known as Cubic, showing that this game is NP-complete even when restricted to only two kinds of blocks.

0.33

Computational Complexity and Exact Algorithms for Pressure-Driven Hydrogen Charging

cs.CC ·2026-10-16 ·Nanase Sato, Wei Wu, Mutsunori Yagiura
theoryunusual
Abstract · 32 pages

We study the problem of deciding the feasibility of pressure-driven hydrogen charging from supply cylinders to demand cylinders without replenishment, asking whether an order of transfers allows every demand cylinder to reach its target pressure. Hydrogen can be transferred from a cylinder to another only when the supply cylinder has higher pressure than the demand cylinder. The charging to each demand cylinder must be completed before starting another, and each transfer continues until the connected cylinders reach equal pressures or the demand cylinder reaches its target. We consider equal-capacity supply cylinders and arbitrary positive rational demand cylinder capacities. Let $n$ be the number of demand cylinders and $L$ the binary input length. With one supply cylinder, feasibility is decidable in polynomial time; with two, it is NP-complete even with cylinder capacities equal to $1$ and nonnegative integer pressures. We prove that feasibility is preserved even if the supply cylinders are always used in non-decreasing order of their current pressures when filling each demand cylinder. We also show that majorization of residual supply cylinder pressure states justifies discarding majorized states in dynamic programming over sets of completed demand cylinders without losing any feasible remaining demand cylinder order. With two supply cylinders, one state per completed demand cylinder set suffices, yielding $2^n\operatorname{poly}(L)$ time and space. With three, the exponential time hypothesis (ETH) excludes deterministic $2^{o(n\log n)}\operatorname{poly}(L)$-time algorithms. This lower bound is tight in the exponent because the rule for order of supply cylinders allows feasibility to be decided in $n!\operatorname{poly}(L)$ time by enumerating only demand cylinder orders.

0.31

Counterexamples to the local antimagic leaf conjecture

cs.DM math.CO ·2026-10-16 ·André Ebling Brondani, Francisca Andrea Macedo França
Abstract · 8 pages, 1 figure

It has been conjectured that $χ_{\mathrm{la}}(T)\leq\ell(T)+2$ for every tree $T$ of order at least three, where $χ_{\mathrm{la}}(T)$ is the local antimagic chromatic number and $\ell(T)$ denotes the number of leaves. We disprove this bound and show that no absolute constant can replace $2$ in an inequality of this form. Our approach uses degree-based intervals containing all possible induced vertex weights under bijective edge labelings, together with a separation criterion that forces selected internal vertex weights to be distinct. We first construct caterpillars $T_3$ and $T_4$ satisfying $χ_{\mathrm{la}}(T_3)=\ell(T_3)+3$ and $χ_{\mathrm{la}}(T_4)=\ell(T_4)+4$, respectively. We then construct, for every positive integer $r$, infinitely many pairwise nonisomorphic caterpillars with exactly $r$ internal vertices for which every bijective edge labeling is antimagic. Each such caterpillar $T$ satisfies $χ_{\mathrm{la}}(T)=|V(T)|=\ell(T)+r$. Consequently, $χ_{\mathrm{la}}(T)-\ell(T)$ is unbounded even within the class of caterpillars, and no absolute constant $C$ can satisfy $χ_{\mathrm{la}}(T)\leq\ell(T)+C$ for every tree $T$.

0.30

Causal Information Exchange as a Necessary Condition for Adaptive Regulation: A Perturbational Framework for Life and Agency

cs.IT ·2026-10-16 ·Brian LuValle, Abicumaran Uthamacumaran, Hector Zenil
theory
Abstract · 59 pages, 30 figures including supplemental sections

Information exchange can support living organisation without providing a sufficient criterion for life. We develop a perturbational framework that distinguishes causal transmission, directional imbalance, internal retention, response structure and functional regulation. Randomised interventions define Shannon information channels; for deterministic binary recipients, their context-conditioned mutual information is exactly a paired perturbation-response indicator. Spatial and temporal averages of these local channels measure information-exposure density, rather than joint information or conserved information flux. Complementary Coding Theorem Method and Block Decomposition Method (BDM) estimates describe response and trajectory structure. A same-shaped zero-mask reference separates raw BDM from a structural contrast; neither is a channel mutual information. Retained exploratory BDM plots are distinguished from the normalised causal-channel results. Retention-weighted coupling and directionality obey algebraic bounds, and ensemble reciprocity need not imply reciprocity within individual contexts. For a specified binary sensor--internal-channel--actuator task, we derive a positive information lower bound from successful regulation and construct controllers with identical information profiles but opposite functional outcomes. A study of 2,000 ordered elementary-cellular-automaton rule pairs, six detailed examples, recurrence and lossless coding analyses, and a signed-network extension demonstrates dissociations among perturbational responsiveness, structural regularity and recurrence. The results establish causal information as a necessary resource for the stated form of adaptive regulation, while showing why neither causal coupling nor response complexity alone measures agency or life.

0.30

Vertex Ordering for Exact Triangle Listing: Concave Geometry and Certified Rounding

cs.DS ·2026-10-16 ·Kourosh Mirsohi, Melika Gorgi, Erfan Hajihashemi
theory
Abstract

Exact forward triangle listers orient each edge of a graph by a vertex order. The order does not change the output, but it changes the work. For the adaptive lister, which scans the smaller of the two forward neighborhoods, the number of candidate tests is a known sum of minima of forward degrees, but how to optimize it had not been characterized. We show that this work is a concave function of the forward degrees, so choosing an order is concave minimization over the orientation polytope, whose vertices are the forward-degree vectors of orders. A single sort therefore rounds any point of the polytope, including the out-degree vector of any orientation, to an order with no more work, with a computable lower bound on the improvement. Minimizing the work is nevertheless NP-hard and APX-hard on triangle-free graphs of maximum degree three, and under the Unique Games Conjecture it has no polynomial-time constant-factor approximation. We also analyze logistic score relaxations and the large-graph limit of degree-based orders; on random cubic graphs, the expected work of every degree-based order is asymptotically at least $2.43$ times the expected optimum. Tangent refinement followed by iterated local search reduces the work by \HeldTIKMean\% on average on nine held-out networks fixed by a pre-registered protocol, and by \NSFrozenTILSMean\% in a supplementary rescoring, specified in advance, against a stronger reference that adds the published Neigh local search. The gains are in candidate tests; we observe no reproducible wall-clock speedup.

0.30

Coopshift: Post-Training Amplifies The Memory Curse In LLM Cooperation

cs.GT cs.CLcs.MA ·2026-10-16 ·Jiayuan Liu, Shiyi Du, Tianqin Li, Emanuel Tewolde, Keyang Xuan +1 more
Abstract

How does post-training change cooperation during repeated interaction? Within each of seven released model lineages, we compare pretrained (Base), supervised fine-tuned (SFT), and subsequent preference- or reward-optimized (Aligned) checkpoints across seven social dilemmas. Copies of each checkpoint play one another for 200 rounds, with visible history limited to the last two or forty rounds. The memory curse refers to lower cooperation when agents can see more past rounds. Our central finding is that post-training amplifies this decline. Averaged across the six binary-action games, expanding visible history reduces cooperation at every SFT and Aligned checkpoint. At each post-trained stage, these declines exceed Base's in six of seven lineages. Most of this average amplification is already present at SFT. With long history, both post-trained stages also cooperate less than Base on average in each binary-action game. In the symmetric binary games, their fitted cooperation rates fall more steeply on average than Base's as per-round cooperation costs rise, benefits fall, or noncooperation becomes more rewarding. The difference in cooperation rates between Aligned and SFT also changes from one-shot decisions to repeated play. When given the same past records, their choices depend on whether the instructions describe a final round or possible future interaction. Post-training evaluation should therefore go beyond isolated choices to assess whether cooperation is sustained in repeated interaction and how it depends on payoffs and visible history.

0.30

Single-Letterization of Rate-Cost Tradeoffs in Static Time-Invariant Control (STIC) Systems

cs.IT math.OC ·2026-10-16 ·Eray Unsal Atay, Venkat Chandrasekaran, Victoria Kostina
Abstract · 38 pages, 9 figures

In settings such as online recommendation platforms, medical decision-making with interim summaries, and stochastic control with static plants, the underlying system is often static so that the next state depends only on the most recent action. Moreover, the true system state is high-dimensional and cannot be observed directly, so actions must be selected based on information received through a rate-limited communication link, which constrains the control performance. This leads to a fundamental rate-cost tradeoff$\unicode{x2014}$the minimum communication rate required to achieve a prescribed average control cost. We study this tradeoff for a class of static systems we term static time-invariant control (STIC) systems, namely, control systems in which the next state depends on the history only through the most recent control action via a time-invariant kernel. For general control systems$\unicode{x2014}$not necessarily static or time-invariant$\unicode{x2014}$our prior work characterized this tradeoff through the limit of a sequence of directed-information minimization problems. However, the dimension of these problems grows with the time horizon, making the limiting formulation computationally intractable. In the present paper, we show for STIC systems that this growing-size formulation admits a single-letter characterization. This identifies STIC systems as the control-side counterpart of Noisy Output is the STate (NOST) channels$\unicode{x2014}$one of the few channel models whose feedback capacity admits a single-letter characterization. This yields, to our knowledge, the first constant-size convex characterization of the rate$\unicode{x2013}$cost tradeoff beyond linear quadratic Gaussian (LQG) control.

0.30

Trembling toward Duverger: Convergence and Cycles in Synchronous Plurality Dynamics

cs.GT ·2026-10-16 ·Maximilian Weigel, Damien Berriaud, Roger Wattenhofer
Abstract · 19 pages, 2 figures

Duverger's law states that plurality elections tend to concentrate support on two candidates. Pivotal-voter models typically derive this from aggregate uncertainty about votes, so that two-candidate competition emerges in the limit of large electorates. We instead propose a model for a fixed, finite population. A poll reveals the current support counts, and each voter best responds under trembling-hand beliefs, assuming that every other voter may deviate with a small probability. Starting from truthful votes, we show that a single update typically suffices for the dynamics to converge to a stable profile in which every voter supports her preferred candidate among the two poll leaders. For any number of candidates and any full-support tremble belief, a four-vote lead of the runner-up over all lower candidates suffices; for three candidates, a lead of two suffices. As a consequence, under random preferences, Duverger's law emerges after one update with probability $1-\mathcal O(N^{-1/2})$. While these results may suggest that the dynamics always settle, convergence can fail. Mild period-two cycles already arise, with a few voters swapping ballots around an otherwise stable profile. We show that substantially richer behavior is possible, even with three candidates. Period-three cycles arise with six voters changing their ballots, and no such cycle can involve fewer than four. Cycles can also alternate the plurality winner. Moreover, larger electorates support ever longer cycles: for every odd $\ell\geq3$, truthful voting can lead to a cycle whose vote counts repeat every $\ell$ rounds and whose full ballot profile repeats every $2\ell$ rounds.

0.29

The Kemeny-Braess Paradox in Random Graphs

cs.DM math.COcs.SImath.PR ·2026-10-16 ·Manuel Bodirsky, Lorenzo Ciardo, Mihyun Kang
Abstract

In 2016, Kirkland and Zeng initiated the mathematical treatment of Braess paradox for random walks on graphs using Kemeny's constant as the measure of travel time. We study this phenomenon for the Erdős-Rényi random graph $G(n,p)$. An edge in a connected graph is called Kemeny-Braess if its deletion preserves connectedness and decreases Kemeny's constant, while a nonedge is Kemeny-Braess if its insertion increases Kemeny's constant. Let $\bar d=(n-1)p$ be the expected degree of $G\sim G(n,p)$, and assume that $\bar d-\log n\to\infty$, so that $G$ is connected with high probability. We determine the sharp asymptotic thresholds for the existence of both types of paradoxical behaviour. With $C^*=1/(1-\log 2)$, the threshold for Kemeny-Braess edges is $\bar d=C^*\log n$, whereas the threshold for Kemeny-Braess nonedges is $\bar d=2C^*\log n$. Hence, as the expected degree $\bar d$ increases, the connected regime exhibits three phases: first Kemeny-Braess paradox occurs for both edges and nonedges, then only for nonedges, and finally it disappears. In particular, in the dense regime, with high probability $G$ has no Kemeny-Braess edges or nonedges. This contrasts sharply with the spectral gap of the normalised Laplacian, for which Braess' paradox persists at constant edge density. Our proof combines a spectral-moment expansion of Kemeny's constant with uniform edge-perturbation estimates. Near the thresholds, the sign of the perturbation is governed by the endpoint degrees, and binomial large deviations yield the constant $C^*$ and the factor of two between the two thresholds.

0.29

Deterministic Global Minimum Cut Using $\tilde{O}(n^{3/2})$ Cut Queries

cs.DS ·2026-10-16 ·Yotam Kenneth-Mordoch
published
Abstract · Accepted at SOSA 2027

Finding a global minimum cut in an unweighted graph is a fundamental task in the cut-query model. While there exists a randomized algorithm for this task using $\tilde O(n)$ queries, known deterministic algorithms use $\tilde{O}(n^{5/3})$ queries. We improve upon this result and give a simple deterministic algorithm that finds a global minimum cut using $\tilde O(\min\left\{ nδ,n^2/δ\right\})$ cut queries, which equals $\tilde{O}(n^{3/2})$ in the worst case, where $δ$ is the minimum degree of the graph. The key ingredient in our approach is an efficient maximum $s,t$-flow algorithm, for the case when the minimum $s,t$-cut has value at most $δ$.

0.28

The Source of Data Value: Consumer Disclosure, Access Design, and Platform Monetization

cs.GT econ.GNmath.OC ·2026-10-16 ·Hexiao Liu, Lijian Lu, Mingzheng Wang
benchmark
Abstract

Digital marketplaces increasingly earn revenue from transaction commissions and seller-facing data services. Sellers may purchase data to increase revenue or avoid competing with an informed rival. We examine how consumer disclosure rights shape these motives and the platform's incentive to provide data. We develop a game-theoretic model of a commissioned platform, two differentiated sellers, and privacy-sensitive consumers. Data generates value through bilateral sales-revenue gains and competition threats arising from asymmetric access. Under mandatory collection, the platform can monetize defensive adoption, even when bilateral personalization reduces seller revenue. Under voluntary disclosure, anonymous purchasing and strictly positive privacy costs make unilateral disclosure unattractive in the baseline. Exclusive access thus yields no usable information, restores sellers' no-data rejection payoffs, and eliminates defensive fees. Profitable provision therefore requires nonnegative bilateral revenue gains. Under full fee extraction, sellers retain their no-data net profits. This revenue alignment does not guarantee welfare gains: additional transactions may benefit consumers, whereas segmentation in an already covered market can raise anonymous prices and reduce welfare. Capping anonymous offers at their no-data benchmark protects individual consumers and limits positive-mass provision to the transaction-creation region. Extensions consider access choice, heterogeneous privacy costs, imperfect identification, and limited contractual credibility. Platforms should evaluate data services based on both bilateral-use and seller-rejection outcomes, and sellers should distinguish revenue-enhancing from defensive investments. Regulators should distinguish equal-access rules from effective consumer control and monitor pricing in anonymous channels.

0.28

Constant-Factor Local Approximation for Bipartite Densest Subgraph

cs.DS ·2026-10-16 ·Michael Kapralov, Jakub Łącki, Silvio Lattanzi, Guy Weissenberg
Abstract

Given a vertex $v$ in a large bipartite graph $G$, can one quickly find a small densely connected region near $v$? This is the algorithmic core of many fraud detection, community detection, and recommender systems. One would like a local algorithm, i.e., an algorithm that probes $G$ in only a small number of locations to find such a region. Our main contribution is the first constant-factor local approximation, independent of the maximum degree of the graph, for the classical notion of bipartite density: if the sides of a bipartite subgraph are $S$ and $T$, its density is $d(S,T):=e(S,T)/\sqrt{|S||T|}$, where $e(S,T)$ is the number of edges between $S$ and $T$. We consider cohesive dense subgraphs, since every dense subgraph contains a cohesive core of at least the same density. For such a cohesive dense subgraph $S\mathbin{\dot\cup}T$ with at most $k$ vertices per side, our algorithm, starting from any vertex in $S\mathbin{\dot\cup}T$, uses $O_ε(k^{4+ε})$ neighbor queries to find a subgraph $X\mathbin{\dot\cup}Y$ of density $Ω_ε(d(S,T))$ on $O_ε(k^{2+ε})$ vertices, for every fixed $ε>0$. Our result improves upon the result of Andersen (2010), whose approximation guarantee depends logarithmically on the maximum degree. We complement our positive result with information-theoretic and computational limitations showing that the quadratic dependence on $k$ in the output size is essentially unavoidable for constant-factor approximation. Our main technical innovation is an adaptation of the Lovász-Simonovits curve to the bipartite density problem, based on weak-$\ell_p$ norm bounds.

0.28

Cooperation Without Adaptation: LLM Agents in a Collective-Risk Dilemma

cs.GT cs.AIcs.CLcs.MA ·2026-10-16 ·Trung-Kiet Huynh, Dao-Sy Duy-Minh, Chi-Nguyen Tran, Phu-Hoa Pham, Phu-Quy Nguyen-Lam +2 more
Abstract · 10 pages, 5 figures, 50-page supplementary material appended

Language-model agents are increasingly deployed to act on behalf of people in settings where individual decisions shape shared outcomes. We examine whether five production models adapt their contributions to changing risks and partner behaviour, or instead follow a relatively fixed strategy. In a collective-risk dilemma game (CRG), six players contribute to a shared pool over ten rounds; if the pool falls short of its target, a catastrophe may destroy their remaining resources. Four models contributed approximately an equal share at every positive risk level, and four of the five also contributed when risk was zero, even though contributing could not improve their payoff. Explicitly stating that only the agent's own payoff mattered reduced zero-risk contributions for most models, while a more neutral prompt did not. Several models correctly identified which action offered the higher expected payoff at different risk levels, yet their contributions changed little. Against fixed partners, no model consistently chose the payoff-maximising response, and in mixed-model groups, a single agent's shortfall in the final round could cause an otherwise successful group to fail. Payoff-based selection also depended on population composition: in a single six-agent group, it favoured the model that contributed least and reduced group success, whereas in larger, randomly mixed populations, it favoured an exact fair-share model at moderate and high risk. These findings show that cooperative self-play alone is not evidence of strategic adaptability. Consistent with a machine-behaviour approach, evaluations of agents in multi-agent systems should test how they respond to incentives, partner actions, and population composition.

0.28

Triangle Covers of Polygons: Approximation Algorithms and Complexity

cs.CG ·2026-10-16 ·Linda Kleist, Lena Scherzer
Abstract · 37 pages, 32 figures, GeoGebra file

We study the geometric set cover problems of covering and partitioning a polygon into a minimum number of triangles. While the decision version of the covering problem is known to be APX-hard and $\exists\mathbb R$-complete (even for polygons without holes and vertices in general position), the computational complexity of partitioning a simple polygon into a minimum number of triangles is open. Our main result refutes a 40 year old conjecture by Asano, Asano, and Pinter stating that the triangle vertices of a minimum partition lie on intersection points of special lines, which are iteratively generated by the polygon vertices and the intersection points. In particular, we show that there exist polygons where all minimum triangle partitions need Steiner points with irrational coordinates. This indicates that the problem might be more complex than previously expected. Complementarily, we provide a simple constant factor approximation to compute a minimum triangle partition of a given polygon (possibly with holes). In the context of coverings, we devise the first constant factor approximation for computing a minimum triangle cover of a simple polygon. For general polygons, an $O(\log n)$-approximation follows by close inspection of previous work.

0.27

On MRD constacyclic codes over inseparable field extensions

cs.IT ·2026-10-16 ·Grégory Berhuy
Abstract

In this paper, we construct MRD constacyclic codes of length $p$ over an inseparable field extension $L/K$, where $K$ has characteristic $p$. In particular, these codes cannot be obtained from known constructions.

0.26

On the Metric Distortion of the Minimum Spanning Tree Problem

cs.DS ·2026-10-16 ·Kian Ghassemi, Masoud Seddighin, Saeed Seddighin
Abstract · 24 pages, 9 figures, 1 table

We study the metric distortion of the minimum spanning tree (MST) problem when numerical edge weights are unavailable and each vertex reports only a ranking of the other vertices induced by these weights. We first show, through a simple example, that without assumptions on the weights, no ordinal algorithm guarantees a finite approximation ratio. We therefore focus on general metrics and present a Borůvka-inspired algorithm using only ordinal comparisons that achieves distortion $O(\log n)$. We complement this result by showing that, for every $\varepsilon>0$, no deterministic ordinal algorithm guarantees distortion at most $3/2-\varepsilon$ on all finite metric instances. We then investigate more informative settings in pursuit of constant distortion, first considering two well-studied metric classes. For tree metrics realized on the given vertex set, a more careful variant of our initial algorithm achieves distortion~$1$, recovering an optimal MST from ordinal information alone. For the broad class of doubling metrics, a different algorithm based on \emph{endpoint cut stability} achieves distortion $2^{O(D)}$, where $D$ is the doubling dimension; this is constant for fixed $D$, independently of $n$. Finally, we study a query-based extension allowing comparisons between two edge lengths without revealing their numerical values. For general metrics with $n\geq2$, fewer than $n$ edge-comparison queries suffice to achieve distortion at most $2\log_2^* n-1$, which is near-constant even for very large inputs.

0.25

Implicit Hitting Set Enumeration for Power Dominating Sets

cs.DS ·2026-10-16 ·Thomas Bläsius, Max Göttlicher, Martin Schirneck
Abstract

The Transversal Hypergraph problem asks to enumerate (list) all inclusion-minimal hitting sets of a hypergraph. It usually assumes that the input is completely known and does not change over the course of the enumeration. In many applications, the underlying hypergraph is too large or expensive to compute. This motivated the implicit hitting set framework [Chandrasekaran, Karp, Moreno-Centeno, and Vempala, SODA 2011] for the corresponding optimization problem of finding a smallest hitting set. Access to the input is given by an oracle that either confirms that a query set $S$ is a hitting set, or reveals a new hyperedge disjoint from $S$. We adapt this framework to enumeration. We show for a general class of algorithms that they are able to find all hitting sets when given access to such an oracle, even if the hypergraph is a priori unknown. We also prove that the current fastest Transversal Hypergraph algorithm, Minimal-to-Maximal Conversion Search by Murakami and Uno [DAM 2014], belongs to this class. We then engineer an implicit hitting set algorithm for the enumeration of minimal power dominating sets in graphs. A power dominating set is a selection of vertices that "observe" all vertices by exhaustively applying two rules: if a vertex is dominated by the selection, it becomes observed; and if an observed vertex has a unique unobserved neighbor, this neighbor becomes observed as well. An experimental evaluation on a collection of power grid instances from the literature shows that our approach achieves very high throughput rates.

0.25

The Strong Total Cover Problem

cs.DM cs.CCmath.CO ·2026-10-16 ·Manuel Aprile, Luca Ferrarini
Abstract · 28 pages, 4 figures

The Strong Total Cover Problem (STCP) is a novel covering problem defined on a graph $G = (V, E)$, where the objective is to find a minimum-cardinality subset $T \subseteq V \cup E$ such that every element of $V \cup E$ is covered. In this setting, a vertex covers itself and all its incident edges, while an edge covers itself together with its two endpoints. The STCP resembles classical covering problems such as Vertex Cover and Edge Cover problems, and naturally arises as the dual of the total matching problem. In this paper, we investigate the STCP from algorithmic, complexity, and polyhedral perspectives. We first establish close connections between STCP, the Dominating Set Problem, and the Maximum Induced Matching Problem. These relationships provide structural insights, imply the NP-hardness of STCP, and identify graph classes for which the problem is polynomial-time solvable. We then investigate the weighted version of STCP, showing that it remains NP-hard even on complete graphs with binary weights. On the other hand, we present a 2-approximation algorithm, and a polynomial-time algorithm for complete bipartite graphs. On the polyhedral side, we introduce the Strong Total Cover Polytope and derive several families of valid inequalities, some of which are shown to define facets. Finally, we provide a complete linear description of the associated polytope for triangular cactus graphs, revealing an interesting contrast with the considerably more intricate facial structure of the dominating set polytope on cactus graphs.

0.25

Pilot-Projected Nonlinear Self-Interference Estimation with Simultaneous Uplink Training

cs.IT ·2026-10-16 ·Seyed Mohammad Azimi-Abarghouyi, Thomas Eriksson
Abstract

This paper considers nonlinear digital self-interference (SI) cancellation and uplink-channel estimation for single-antenna in-band full-duplex radios when a desired uplink pilot is present during SI training. In this setting, a rich nonlinear SI dictionary can absorb part of the structured pilot response, while its large and correlated set of basis functions can make reliable SI estimation difficult, particularly with limited training. During SI-coefficient estimation, we treat the unknown uplink channel as a nuisance parameter and eliminate it to obtain an equivalent reduced SI estimation problem on the orthogonal complement of the pilot convolution subspace. Since the projected nonlinear dictionary can remain ill conditioned, we develop two complementary full-dictionary estimators: a pilot-projected ridge estimator with Bayesian-information-criterion tuning (PP-RidgeBIC), which uses global ridge shrinkage, and a pilot-projected group-sparse Bayesian estimator (PP-GSBE), which learns branch-dependent Bayesian shrinkage from the data. These estimators offer two complementary operating points: PP-RidgeBIC favors computational efficiency, whereas PP-GSBE provides greater adaptivity to heterogeneous nonlinear SI components. Simulations show substantial gains in SI cancellation and uplink reception over non-projected and hard-selection alternatives; notably, PP-GSBE suppresses the residual SI below the noise floor with moderate training and consistently provides the strongest performance under nonlinear model mismatch.

0.24

Achievable Rate Region Characterization of Rotatable Antenna-assisted Interference Channel

cs.IT ·2026-10-16 ·Junran Pei, Guangji Chen, Qingqing Wu, Qiaoyan Peng, Wei Cai
Abstract

This paper characterizes the achievable rate region of a rotatable antenna (RA)-assisted multiple-input single-output (MISO) interference channel. We exploit the additional spatial degrees of freedom (DoFs) offered by the dynamic orientation of RAs to simultaneously enhance the desired signal and suppress interference, thereby significantly enlarging the achievable rate region of the interference network. To this end, we invoke the rate-profile technique to formulate a sum-rate maximization problem for jointly optimizing the transmit beamforming and the array rotation, subject to the practical mechanical constraints of the RAs. For the special case of a two-receiver scenario, we introduce the concept of subspace correlation coefficient to derive closed-form solutions for the optimal RA rotation angles and transmit beamforming vectors. Inspired by the insights obtained in the two-receiver special case, we propose an efficient two-step optimization framework to obtain a high-quality solution for the general multi-receiver case. In particular, the array rotation angle of each transmitter is optimized to minimize the leakage interference to non-associated receivers. With the optimized array rotation angles, the optimal transmit beamforming is obtained by invoking the second order cone programming based approach. Numerical results demonstrate that the proposed solution achieves near-optimal performance and that integrating RAs effectively simplifies transceiver designs by enabling simple maximum ratio transmission (MRT) beamforming with negligible performance loss.

0.24

Complexity of liveness in ring Petri nets and of halting in chip-firing on simple digraphs

cs.CC cs.DMcs.FL ·2026-10-16 ·Eike Best, Raymond Devillers, Petr Jančar, Matěj Ošťádal
Abstract

A ring Petri net (weighted circuit, cyclic weighted T-system) consists of a single directed circuit with arbitrary positive arc weights. We show that deciding liveness of ring Petri nets is coNP-complete, and that coNP-hardness holds in the strong sense already for token-conserving rings, in which every transition consumes and produces the same number of tokens. The proof is a short reduction from the covering problem for systems of residue classes (the complement of Simultaneous Incongruences). Together with a known reduction to circuits, it follows that liveness of weighted T-systems (weighted marked graphs, synchronous dataflow graphs) is coNP-complete, and coNP-hard in the strong sense. As a further consequence, the halting problem for chip-firing on simple directed graphs is NP-complete, which answers a question of Farrell and Levine. On 7 October 2026, the problem of liveness was posed to the AI model Claude Opus 5.5 (Anthropic), which found the reduction, also suggesting the application to the chip-firing problem, and wrote a first version of this paper; the human authors checked it and, together with the model, revised it iteratively into the present form.

0.24

Gaussian Identification in MIMO Channels

cs.IT ·2026-10-16 ·Mohammad Javad Salariseddigh, Christian Deppe
Abstract · 43 pages, 1 figure

We study the identification capacity of discrete-time Gaussian multiple-input multiple-output (MIMO) channels under per-antenna, per-symbol peak-amplitude constraints. We establish upper and lower bounds on the identification capacity that scale linearly with the channel rank $r.$ The number of reliably identifiable messages can grow super-exponentially with blocklength $n,$ following $M_n(R)=2^{(n\log n)R},$ where $R$ is an achievable rate. The channel rank provides a spatial-multiplexing advantage, enabling higher achievable rates and, consequently, stronger super-exponential codebook growth. These results characterize the rank-dependent gain in achievable identification performance.

0.23

Critical-state algorithms for lot sizing with all-units discounts and variable inventory bounds

cs.DS ·2026-10-16 ·Kleitos Papadopoulos
theory
Abstract

We study deterministic single-item lot sizing with all-units quantity discounts, linear holding costs, and time-varying end-of-period inventory bounds. Unit prices may vary arbitrarily over time. For one common positive discount threshold, we give an exact algorithm requiring $O(n^3)$ arithmetic operations and $O(n^2)$ working space, including recovery of an optimal purchase plan. The method combines a critical-state theorem in cumulative-purchase coordinates with two prefix-minimum scans per dynamic-programming layer. The structural argument accommodates downward price discontinuities and distinguishes prescribed from free terminal inventory. With a fixed global set of $m$ positive threshold values, the bounds become $O_m(n^{m+2})$ time and $O_m(n^{m+1})$ space. Period-specific thresholds admit the same exact construction, but its state set can be exponential. For commensurate flow data, an inventory-lattice implementation takes $O(PW)$ time, where $P$ is the total number of price tiers and $W$ is the largest inventory-lattice width. We also derive a conditional optimality certificate from a monotone one-threshold relaxation; this certificate does not imply a near-linear worst-case algorithm for arbitrary prices. Exact-arithmetic implementations are checked against exhaustive inventory dynamic programs and tier-enumeration linear programs. An illustrative timing study identifies settings in which the lattice implementation is preferable.

0.23

New lower bounds for binary constant-weight codes from prescribed-automorphism search, and two quantum codes meeting their upper bounds

cs.IT math.CO ·2026-10-16 ·Chris Dzoba
theory
Abstract · 12 pages. Code word lists, group generators, orbit representatives, two independent checkers, verification outputs, quantum stabilizers and …

We give explicit binary constant-weight codes that raise the lower bounds on $A(n,6,w)$ in eight cells of Brouwer's table, all now listed there. The first round (9 October 2026) gave four values: $A(27,6,11)\ge 19305$, $A(27,6,12)\ge 24453$, $A(28,6,10)\ge 18642$ and $A(28,6,9)\ge 11154$, together with an explicit $A(24,6,12)$ code of 5566 words, now in the table's list of lost codes. The second round (14 October 2026) gave seven values, listed in the table since 15 October 2026: $A(29,6,10)\ge 27183$, $A(29,6,11)\ge 42042$, $A(29,6,9)\ge 14781$, $A(28,6,10)\ge 19110$, $A(27,6,12)\ge 24804$, $A(28,6,9)\ge 11388$ and $A(27,6,13)\ge 27783$, three of which replace first-round values. Every code is a union of orbits of the semilinear affine group $\mathrm{AΓL}(1,27)$, its index-2 subgroup, or $\mathrm{AGL}(1,23)$, acting on all but at most two coordinates. The orbits were chosen by maximum-weight independent set search on the orbit conflict graph: branch and bound in the first round, iterated local search in the second. Each code was checked by two separately written programs over all pairs of words and was regenerated from its generators and orbit representatives. We also report two qubit stabilizer codes, $[[20,11,4]]$ and $[[43,32,4]]$, which raise the lower bounds of 3 listed at codetables.de to the listed upper bound 4.

0.23

A Class of Rényi-based Concentration Metrics for Kolmogorov-Arnold Geometry in Multilayer Perceptrons

cs.IT stat.ML ·2026-10-16 ·Juan Carlos Fernández, Asunción Jiménez-Cordero, José Francisco Paris
Abstract

The interpretability of artificial intelligence models has become a central concern, especially for black-box models such as multilayer perceptrons (MLPs). Kolmogorov-Arnold Geometry has recently emerged as a framework for analysing the internal representations of these networks, by examining how the hidden layer locally responds to changes in the input through the minors of its Jacobian. The degree to which a few of these minors dominate the rest, known as their concentration, has so far been measured with the Participation Ratio. Although this metric has useful properties, its maximum value depends on the size of the network, so a given value cannot be read as high or low concentration without reference to the architecture. To overcome this limitation, in this paper we propose the $α$-RC metrics, a family of Rényi-based concentration measures that compare the normalized magnitudes of the minors with the uniform distribution through the Rényi divergence of order $α$. All the metrics in the family are bounded between 0 and 1 regardless of the architecture, with 0 indicating that the magnitude is evenly spread over the minors and 1 that it is carried by a single one. Moreover, the metrics are non-decreasing in $α$, and for $α$ = 2 they reduce to an explicit transformation of the Participation Ratio. Computational experiments with single-hidden-layer MLPs trained on a synthetic dataset show the benefits of our proposal.

0.23

Explicit construction of optimal binary linear complementary pairs via Belov codes

cs.IT ·2026-10-16 ·Yansheng Wu, Jing Chen, Jong Yoon Hyun
Abstract · 30 pages

The study of linear complementary pairs (LCPs) has been a hot topic in coding theory. Recently, Güneri constructed an infinite family of optimal binary LCPs from Solomon-Stiffler codes. Specifically, he proved that for every $k \geq 5$ and $d \geq \lceil (k - 1)/2 \rceil 2^{k - 1}$, there exist optimal binary linear complementary pairs of codes of length $g_2(k,d)$, where $g_2(k,d)$ denotes the Griesmer bound. In this paper, we construct additional optimal binary LCPs from Belov codes, which form a generalized class of Solomon-Stiffler codes. Our results not only generalize Güneri's construction, but also enable more flexible construction of optimal binary LCPs with a wide range of minimum distances.

0.22

Dynamic Load Balancing under Restricting Updates

cs.DS ·2026-10-16 ·Aaron Bernstein, Sayan Bhattacharya, Peter Kiss, Thatchaphol Saranurak
Abstract

We study the problem of load balancing in the dynamic setting. The input is a bipartite graph $G = (J \cup M, E)$ of jobs $J$ and machines $M$. Every job has a positive weight $w(j)$. The goal is to assign every job in $J$ to an adjacent machine in $M$ so as to minimize the maximum machine load, where the load of a machine is the sum of the weights of the jobs assigned to it. Although load balancing has been studied extensively in many models of computation, the only previous result in the dynamic model addresses the \emph{incremental} setting, where updates take the form of edge insertions and job deletions [BKS23b]. The authors show a $(1+ε)$-approximation for incremental \emph{fractional} load balancing with $\text{poly}(\log(n),1/ε)$ update time; they point out that their techniques do not extend to the decremental setting. Main Result: We obtain an algorithm which maintains a $(1+ε)$-approximate solution to the fractional load balancing problem under edge deletions and job insertions in $\tilde{O}(\log (n), 1/ε)$ update time. Our techniques are very different from those of the incremental algorithm of [BKS23b]. Instead, our algorithm is inspired by a load-balancing algorithm from the {\sc congest} model of computation [ALPZ21]. We develop a novel analytic framework for this algorithm, which allows us to show that it performs well in the dynamic setting. Rounding Scheme: We also present a black-box rounding scheme for dynamic load balancing. Plugging in our fractional algorithm, we obtain a decremental algorithm for $(2+ε)$-approximate \emph{integral} load balancing with $\text{poly}(\log(n),1/ε)$ update time. We also obtain the same bounds for incremental load balancing by plugging in the fractional algorithm of [BKS23b].

0.22

Dimension Thresholds for Relative-Error Representations of Finite Metrics in Pseudo-Euclidean Spaces

cs.CG math.MG ·2026-10-16 ·Nikita Muravya, Aleksandr Bespalov, Anna Vozianova
theory
Abstract

A finite metric space with $N$ points admits an exact representation in a pseudo-Euclidean space of dimension at most $N-1$. We study the minimum dimension required for relative-error approximation of squared distances. Our main result gives an upper bound for arbitrary metrics in terms of local distance ratios: for $4/5<\varepsilon<1$, dimension $O((N/k)\log N+kρ_k^4\log N/(\varepsilon-4/5)^2)$ suffices, where $ρ_k$ is the largest ratio, over all points, between the distance to the $k$th closest point (counting the point itself) and the nearest-neighbour distance. The proof combines a low-rank landmark approximation with a sparse correction controlled by its factorization norm. For metrics with distances in $[1,Φ]$, where $Φ\ge5$ is fixed, the worst-case dimension is $Θ(N)$ for $0<\varepsilon<4/5$, lies between $Ω(\sqrt N)$ and $O(\sqrt N\log N)$ for $4/5<\varepsilon<12/13$, and is $Θ(\log N)$ for $\varepsilon>(Φ^2-1)/(Φ^2+1)$. For $Φ>5$, an intermediate range remains without matching bounds. Sign-pattern counting yields the $Ω(\sqrt N)$ lower bound for $4/5<\varepsilon<12/13$, while Hadamard matrices give explicit linear lower bounds for $\varepsilon<4/5$ via a known approximate-rank estimate. We also construct configurations with null squared intervals whose minimum representation dimension remains $N-1$ for every $\varepsilon<1$. Finally, for every fixed $0<\varepsilon<3/5$, we show that the quadratic dependence on the parameter $C$ in the dimension bound derived from Deng et al.'s pseudo-Euclidean Johnson-Lindenstrauss lemma is necessary up to a logarithmic factor.

0.22

Power of Conditional Sampling on Hypergraphs

cs.DS ·2026-10-16 ·Debarshi Chanda, Buddha Dev Das, Arijit Ghosh, Gopinath Mishra
theory
Abstract

Let $G=(V,E)$ be a $d$-uniform hypergraph with $|V|=n$ and $|E|=m$. Estimating $m$ from uniformly random hyperedge samples requires $Θ(\sqrt{m})$ samples by a birthday-paradox argument. We investigate the power of conditional sampling through a hierarchy of oracles: for an integer $1\le s\le d$, given pairwise disjoint subsets $X_1,\ldots,X_s\subseteq V$, the $s$-colorful oracle returns a hyperedge chosen uniformly at random from all hyperedges contained in $X_1\cup\cdots\cup X_s$ that intersect every $X_i$, provided one exists. For $s=1$, this samples a uniformly random hyperedge induced by $X_1$. For every $1\le s\le d$, we give a randomized algorithm that outputs $\widehat{m}\in[(1-\varepsilon)m,(1+\varepsilon)m]$ with probability at least $2/3$ using $O_{s,d}(\operatorname{poly}(1/\varepsilon,\log n))$ conditional samples, where the exponent of $\log n$ is an absolute constant independent of both $s$ and $d$. Conditional sampling thus yields an exponential separation over uniform hyperedge sampling. We also study the decision version, in which the oracle only reports whether a qualifying hyperedge exists. For every $1\le s<d$, we prove that any algorithm using the $s$-colorful decision oracle requires $\widetildeΩ_{s,d}(n^{\frac{d-s}{d-s+1}})$ queries. Hence, for $s<d$, access to uniformly random witness hyperedges (random certificates) is exponentially more powerful than existence queries. For $s=d$, decision queries alone achieve $\operatorname{poly}(1/\varepsilon,\log n)$ query complexity, but the exponent of $\log n$ is known to be $Θ(d)$. The colorful hierarchy thus exhibits a phase transition at $s=d$: for $s<d$, random certificates are essential for polylogarithmic-query estimation, whereas at $s=d$ decision queries suffice, yet random certificates still provably reduce the exponent of $\log n$ from $Θ(d)$ to an absolute constant.

0.21

Visibility Sets, Empty Pentagons, and Sufficient Conditions for Hexagons

cs.CG math.CO ·2026-10-16 ·Carlos M. Nicolas
theory
Abstract · 21 pages, 12 figures

Given a set $S$ of $n$ points in general position in the plane, we define the $C$-set $C(p,q)$ of an oriented edge $(p,q)$ as the set of extreme points of the visibility set $V(p,q)$ facing $(p,q)$. Since every point of $V(p,q)$ that is not in $C(p,q)$ gives rise to an empty convex pentagon having $(p,q)$ as a boundary edge, it is natural to study the properties of $C$-sets in order to bound the number of empty convex pentagons of $S$. We prove tight bounds on the intersection size and union complexity of $C$-sets sharing an endpoint. We also consider the problem of bounding the number of empty pentagons and hexagons in terms of the number of interior and boundary points of $S$. While the hexagon bound is weak, it offers a potential step toward a superlinear lower bound. On the other hand, the two-parameter bound for pentagons, combined with the structural results on $C$-sets, yields a simple proof of an $Ω(n^{3/2})$ bound for the number of pentagons. While this is a significant improvement over previously published bounds, a recent preprint \cite{Astudillo2026} attains a better bound using completely different methods. Thus, the primary contribution of this paper lies in the structural results about $C$-sets and the two-parameter bounds for empty pentagons and hexagons.

0.21

Rapid mixing within a phase for Ising models

cs.DS math.PRmath-ph ·2026-10-16 ·Vasyl Paliy, Youngtak Sohn
theory
Abstract · 42 pages

We study Glauber dynamics for Ising measures conditioned to have a sufficiently large majority of $+1$ spins. Under a sparse spectral width condition on the interaction matrix, we derive a Poincaré inequality for the conditioned measures from a magnetization drift condition, and a modified logarithmic Sobolev inequality from a stronger, entropy weighted condition. As applications, we obtain approximate posterior sampling in total variation distance for $\mathbb{Z}_2$ synchronization at high enough signal-to-noise ratio using $O(n\log n)$ Glauber updates from a spectral initialization. For ferromagnetic Ising models on random $d$-regular graphs, we prove rapid phase ordering down to inverse temperatures of order $1/d$, extending the temperature range of Gheissari, Sly, and Sohn (2025). Finally, we extend these inequalities to unconditioned Ising measures under an inward drift condition outside the conditioned region. This yields $O(n\log n)$ mixing of Glauber dynamics for the Sherrington-Kirkpatrick model at every fixed inverse temperature when the external field is sufficiently large.

0.20

Approximating the Girth of an Undirected, Weighted Graph for Larger Approximation Factors

cs.DS ·2026-10-16 ·Eliyahu Yablon
Abstract

The girth of a weighted undirected graph is defined as the length of a shortest cycle in the graph. In this paper, we give a randomized $O\left(m^{1+2/(k-1)} n^{1/(k-1)}\right)$ algorithm to approximate the girth of a graph with $n$ vertices and $m$ edges to within a factor of $k$ for any real $k$. We assume that all edge weights are non-negative. For sufficiently high approximation factors, our algorithm outperforms the state of the art by polylogarithmic factors. Specifically, for approximation factors $k = c\frac{\log n}{\log \log n}$ where $c>5$, our algorithm gives a runtime of $O\left(m (\log n)^{5/c}\right)$, while instantiating the previous best $k$-approximation algorithms with $k = c\frac{\log n}{\log \log n}$ yields a running time of $O\left( \frac{cn (\log n)^{2 + 4/(3c)} }{\log \log n} + m\log n\right)$ (Kadria, Roditty, and Vassilevska Williams, arXiv:2607.00938). Furthermore, the algorithm presented here is conceptually simple, has relatively small constants, eschews single-source shortest path and sparse spanner algorithms, and employs a new technique, which we call ``random filtering.'' That being said, the basic algorithm returns only a numerical estimate for the girth, rather than a witness cycle.

0.18

Four New Single-Error-Correcting Binary Codes

cs.IT math.CO ·2026-10-16 ·Moshe Milshtein
Abstract

We construct a $(16, 3072, 3)$-code, a $(17,6144,3)$-code, an $(18,12288,3)$-code and a $(19,24576,3)$-code. Consequently, we get four new lower bounds: $A(16,3) \geq 3072$, $A(17,3) \geq 6144$, $A(18,3) \geq 12288$ and $A(19,3) \geq~24576$. Each of the codes consists of a union of cosets of a linear code. The constructions are the result of a computer search executing a tabu search algorithm.

0.17

Unconditional Quantum PAC Learning Advantage in the Large Error Regime

cs.CC quant-ph ·2026-10-16 ·Igor C. Oliveira
published
Abstract

We establish an unconditional super-polynomial separation between quantum and classical PAC learning of Boolean functions under the uniform distribution in the large error regime. For every advantage $λ(n) = n^{-ω(1)}$, we construct a concept class that a polynomial-time quantum learner learns to agreement $1/2+λ(n)$ with success probability at least $1-1/n^k$. The learner uses only quantum random examples and outputs a polynomial-size deterministic classical circuit. In contrast, no polynomial-time classical learner achieves agreement $1/2 + λ(n)/n^k$ with success probability greater than $1/n^k$, even with membership queries. To achieve this, we develop a nearly-optimal quantum learner that uses quantum random examples to produce a hypothesis achieving a prescribed correlation with an arbitrary Boolean function. The learner generalizes Fourier sampling from parity hypotheses to quadratic hypotheses. While standard Fourier sampling is insensitive to the sign of the correlation, we use interference to bias sampling toward positively correlated hypotheses. This is combined with a careful analysis of the resulting distribution to boost both the learner's success probability and its agreement with the target function. We believe that the resulting quantum learning algorithm might be of independent interest. Finally, although a robust quantum PAC learning advantage holds unconditionally in the large error regime, we note that an analogous separation between quantum and classical learning in the small error regime would imply that $\mathsf{NP} \nsubseteq \mathsf{BPP}$.

0.15

Online Graph Coloring with Random Arrivals

cs.DS cs.DM ·2026-10-16 ·Hirotaka Yoneda, Masataka Yoneda
theory
Abstract

We study online graph coloring in the random-arrival model, where the vertices are presented in a uniformly random order. Assuming $n$ and $k$ are known in advance, we present a randomized exponential-time algorithm that colors any $k$-colorable graph on $n$ vertices with $O(k^3 \log^2 k \log^4 n)$ colors in expectation. This is the first online coloring algorithm that uses a polylogarithmic number of colors for all fixed $k \geq 3$, and is better than a known lower bound of $Ω(\log^{k-1} n)$ colors under adversarial arrivals for all fixed $k \geq 6$. In addition, assuming $n$ is known in advance, we obtain a competitive ratio of $\widetilde{O}(n^{2/3})$. For comparison, it is known to be NP-hard to approximate (offline) graph coloring within a factor of $n^{1-\varepsilon}$ for all fixed $\varepsilon > 0$. Therefore, our exponential-time online algorithm achieves a performance guarantee beyond the NP-hardness threshold for polynomial-time approximation algorithms.

0.14

Bayesian Learning of Parametric Mean Field Games with Common Noise

cs.GT ·2026-10-16 ·Grégoire Lambrecht, Mathieu Laurière
Abstract

We study the inverse problem of recovering unknown parameters governing the dynamics and rewards of Mean Field Games (MFGs) from partial observations of population trajectories, made stochastic by common noise. The observer seeks to recover a consistent parameter $θ$ via Bayesian inference, facing the challenge that evaluating the likelihood requires solving a new Nash equilibrium (NE) for each candidate $θ$. Due to common noise, the mean-field evolution is stochastic and equilibrium policies are population dependent, which makes solving for each parameter computationally expensive. To overcome this, we introduce the Bayesian-Extended Mean Field Game (BE-MFG), whose single NE simultaneously encodes equilibrium policies for all parameters, enabling efficient learning of the conditional density of mean-field trajectories. We formalize parameter consistency and identifiability in MFGs with common noise and investigate conditions under which consistent parameter recovery is guaranteed. On three synthetic environments of various complexity, we recover the ground-truth parameter accurately from a small number of partial, stochastic observations, outperforming an adaptation of inverse reinforcement learning for MFGs. A fourth experiment, on public electricity market data, suggests that the approach can also be applied to real-world data.

0.12

Computing Multi-Slot First-Price Pacing Equilibria at Scale via Entropic Smoothing

cs.GT math.OC ·2026-10-16 ·Yuchen Xie, Shenyinying Tu, Yuan Gao, Yi Zhang, Deepak Kumar
Abstract · 40 pages, 6 figures, 5 tables

Pacing equilibria describe how budget-constrained advertisers clear an ad market, but computing one at production scale is hard when every request fills several slots: the allocation jumps whenever two campaigns' scores cross. We consider the first-price pacing equilibrium (FPPE) of auctions with $K$ homogeneous slots and unit-demand campaigns, whose Eisenberg-Gale dual reduces, via a TopK identity, to a convex program over bids. Regularizing each request's allocation with the binary (Fermi-Dirac) entropy gives a smooth, strictly convex dual whose gradient is exactly spend minus budget and whose evaluation splits into independent one-dimensional threshold solves, one per request. The smoothed equilibrium converges to the hard one at an explicit $O(\sqrt{T})$ rate, and a duality gap certifies how far any bids are from the hard equilibrium. We implement an efficient multi-GPU solver and test it on a production ad workload. We use it to compute an equilibrium price anchor: each campaign's hindsight equilibrium cost per key result.

0.12

Electrostatic Limits on Quantum Spatial Search

cs.CC quant-ph ·2026-10-16 ·Weikang Li
theory
Abstract

We prove that a quantum walker cannot search a two-dimensional lattice in Grover's $O(\sqrt N)$ time, resolving the long-standing open question posed by Aaronson and Ambainis in 2003. A walker that hops between neighbouring sites of an $L\times L$ lattice, $N=L^2$, needs $Ω(\sqrt{N\log N})$ local steps at least to find a marked vertex, even with unlimited memory and free queries. Hence, Tulsi's algorithm is asymptotically optimal. Our proof rests on a surprising correspondence with classical electrostatics. We treat the marked vertex as a point charge and weight the overlap between the walker's states with and without the mark by the potential drop, which vanishes where queries act. Only local steps change this weighted overlap, at a rate limited by the field energy, so the lower bound is set by the self-potential of the charge. In the plane this self-potential grows logarithmically with system size, like the Coulomb potential, leading to the extra root-logarithmic factor. Several marked vertices turn the charge into a conductor, and the lower bound depends on their arrangement through its capacitance.

0.11

Concrete Functions in Private Simultaneous Quantum Messages: Quantum Advantages and Structural Limitations

cs.CC quant-phcs.CR ·2026-10-16 ·Harumichi Nishimura
theory
Abstract · 45 pages

Private simultaneous quantum messages (PSQM) provide a minimal model of noninteractive secure multiparty quantum computation, yet their communication complexity for concrete functions remains poorly understood. We show that familiar total functions already exhibit both genuine quantum advantages and strong structural limitations. For three-party equality on $n$-bit inputs, which determines whether the players' $n$-bit strings are identical, we construct an entanglement-assisted protocol with asymptotic communication $(1.633\ldots)n$ qubits, whereas every perfectly correct protocol without prior entanglement requires at least $3n$ qubits. For three-bit majority, one shared GHZ state and one transmitted qubit per player suffice for perfect correctness and perfect privacy, reducing the communication from the optimal classical cost of six bits to three qubits. Three-bit AND behaves differently. No perfectly correct and perfectly private protocol exists when the players share an arbitrary three-qubit state, apply arbitrary local unitary encodings, and transmit one qubit each. However, increasing the local dimension overcomes this obstruction: one shared three-qutrit GHZ state and one transmitted qutrit per player yield a one-shot protocol with total communication $3\log_2 3$. Together with the six-bit classical lower bound, this gives a strict quantum communication advantage. More generally, for every $k\geq 3$, we construct a perfectly correct and perfectly private one-shot protocol for $k$-bit AND using one qudit of dimension at most $k$ per player, and hence at most $k\log_2 k$ qubits in total.

0.11

From Border-Agnosticism to Causal Rewrite Systems

cs.DM cs.DC ·2026-10-16 ·Pablo Arrighi, Marin Costes, Luidnel Maignan
Abstract

Causal Graph Rewriting was recently introduced as an evolution of Causal Graph Dynamics to allow for a controlled form of asynchronism in these dynamical systems with dynamical graphs. In this context, the notion of borders---akin to dangling edges---has gained increasing importance. In this paper, we re-examine these borders, interpreting them as missing information. This perspective adds structural rigor to the framework, allowing border-agnostic causal graph rewriting to be reformulated as a rewriting system rather than a dynamical system. This shift represents a crucial step toward facilitating the treatment of reversibility.

0.10

An Erdős-Pósa theorem for induced packings of even cycles

cs.DM math.CO ·2026-10-16 ·Jungho Ahn, Lars Jaffke, O-joung Kwon, Paloma T. de Lima, Hyemi Park
theory
Abstract · 38 pages, 5 figures

We show that for every graph $G$ and every positive integer $k$, $G$ contains either $k$ vertex-disjoint even cycles with no edges between distinct cycles, or a set of $O(k^2\log k)$ vertices whose closed neighborhood meets every even cycle of $G$. Moreover, we give a polynomial-time algorithm that outputs one of the two outcomes. As an application, we show that every $K_{1,t}$-free graph containing no induced packing of $k$ even cycles has tree-independence number $O(tk^2\log k)$, with a corresponding decomposition computable in polynomial time.

0.10

On the Convergence of Iterated Round Robin

cs.GT ·2026-10-16 ·Maxime Lucet, Henri Surugue, Nawal Benabbou, Aurélie Beynier, Nicolas Maudet +1 more
Abstract

Round Robin (RR) is a simple and widely used picking sequence mechanism for allocating indivisible goods, known to guarantee envy-freeness up to one item (EF1) when preferences are additive. Despite being manipulable (meaning that agents may have an incentive to strategically misreport their preferences), it is known to always admit at least one pure Nash equilibrium. In this paper, we study an iterated variant of RR, in which agents observe the outcome of the mechanism at each iteration and may revise their reported preferences through a best-response dynamics. We first consider a full-information setting, where agents observe both the outcome and the reported preferences of all the agents. In this setting, we show that cycles can occur even under strong restrictions on the manipulations carried out by the agents. We nonetheless identify a natural restriction (item-preserving swaps), under which convergence to a pure Nash equilibrium is guaranteed to exist for any number of agents and items. We also show that restricting to adjacent swaps guarantees convergence for two agents and a limited number of items, leaving the question open for unbounded number of items. We then turn to a partial-information setting, in which agents only observe the resulting allocation and must then infer the preference profiles of the others assuming consistency with the RR sequence. We study three behaviours guiding the agents when deciding how to manipulate under uncertainty on the other agents' preferences. We show that two of the studied behaviours, despite being quite natural, may cycle while a more cautious rule guarantees iterated RR to be strategyproof.

0.10

A Generalized Half-Singleton Bound for List Decoding of Insertion/Deletion Errors

cs.IT math.CO ·2026-10-16 ·Neehar Verma, Camilla Hollanti, Razane Tajeddine
theory
Abstract

We derive an upper bound on the rate of a linear code capable of list decoding from insertion and deletion errors. Subsequently, we show that a linear code can list decode from at most $n - \frac{L}{L+1}k + 1$ insdel errors with fixed list size $L$, and that this bound is strict when the code does not contain the all-ones vector. We also provide a simple construction of codes achieving the optimal rate for any given list size capable of correcting a single insdel error. All the results hold for linear codes over finite fields of size $q \geq 3$.

0.07

Envy-Free Randomized Apportionment

cs.GT ·2026-10-16 ·Francesco La Commare, Jamie Tucker-Foltz, Victor Verdugo
theory
Abstract

The United States allocates representatives to each state in proportion to its population, while many parliamentary democracies select representatives from party lists in proportion to the number of votes each party receives. In both settings, the legislature's small size necessitates rounding to the nearest whole number of seats. Recent work has shown a plethora of ways to randomly round seat shares to achieve perfect ex ante fairness. As a result, party (state) $i$ may have a larger number of votes (population) than party $j$ but receive fewer seats. In this case, we say that party $i$ envies party $j$. We investigate the extent to which such envy is avoidable. Specifically, we seek to design apportionment rules that minimize the expected number of envious pairs $(i, j)$. If the expectation is zero, we say that the seat allocation is envy-free. On the negative side, we show that any $n$-party apportionment rule that ignores lower quotas (the floors of the deserved number of seats for each party) suffers from a worst-case envy of $Θ(n^2)$, even on instances where an envy-free allocation exists. All previously studied apportionment rules fall into this class and thus do a poor job of minimizing envy. In the case where all lower quotas are equal, we prove that all allocations have the same expected envy. In the general case, we we introduce a general polynomial-time LP framework for a large class of apportionment objectives, including minimizing the expectation and variance of envy, by efficiently solving the separation problem on the dual. Finally, we empirically demonstrate that envy-free allocations exist very frequently, including in every historical instance from the US decennial census. Taken together, these results provide a clear recipe for implementing randomized apportionment in a less objectionable way in practice.

0.06

Revisiting the Fairness of Pure Nash Equilibria of Round-Robin

cs.GT ·2026-10-16 ·Georgios Amanatidis, Haris Aziz, Georgios Birmpas, Federico Fusco, Stefano Leonardi +1 more
theory
Abstract · Corrects the claim in arXiv:2109.08644 (WINE 2021; Math. Oper. Res. 49(4), 2024) that every pure Nash equilibrium of Round-Robin is EF1. We …

We prove that for additive valuations, not every pure Nash equilibrium of the Round-Robin mechanism induces an allocation that is envy-free up to one good (EF1) with respect to the agents' true valuations. This corrects an earlier claim of Amanatidis et al. (WINE 2021; Math. Oper. Res. 2024). On the positive side, every pure Nash equilibrium of every fixed recursively balanced picking sequence mechanism induces an allocation that is $1/2$-EF1 and proportional up to one good (PROP1). We prove the approximate EF1 guarantee through a potential function comparing two executions, and the proportionality guarantee via a deviation that preserves the agent's most valuable goods. We also give a family of Round Robin equilibria showing that it is not possible to have a strictly better guarantee than $2/3$-EF1 for all equilibria.

0.06

Robotics, Control & Signals 102

QIEN: Quantum-Inspired Energy-Neutral UAV Trajectory for WPT-Enabled Industrial HRLLC

eess.SY ·2026-10-16 ·Ali Ranjha, Georges Kaddoum, Halim Yanikomeroglu
publishedbenchmark
Abstract · 14 pages. Accepted for publication in IEEE Transactions on Vehicular Technology. This work was supported by the Postdoctoral Research Fellow…

This paper introduces Quantum-Inspired Energy-Neutral (QIEN) UAV Trajectory for WPT-Enabled Industrial HRLLC, a novel framework for mobility-integrated trajectory and resource optimization in solar-powered uncrewed aerial vehicle (UAV) relay systems supporting autonomous mining robots in post-disaster environments. We formulate a mixed-integer nonlinear program (MINLP) that jointly optimizes UAV trajectory, finite blocklength allocation, transmit power, and wireless power transfer (WPT) over discrete time horizons to maximize aggregate data throughput. The system operates under strict extreme hyper-reliable low-latency communication (HRLLC) constraints, energy sustainability requirements via solar harvesting, and continuous robot power delivery. To solve this large-scale optimization problem, QIEN employs a hybrid Cooperative Quantum-behaved Particle Swarm Optimization (Co-QPSO) algorithm using quantum tunneling mechanisms and a heterogeneous swarm architecture. Our simulation results demonstrate that the proposed QIEN framework significantly outperforms benchmark algorithms, achieving successful convergence while maintaining energy-neutral operation and satisfying all reliability, power, and mobility constraints.

0.65

NestedVLA: Learning to Consolidate and Generate Skills for Vision-Language-Action Models

cs.RO cs.LG ·2026-10-16 ·Guodong Du, Wanyu Lin
top venuecode/data
Abstract · Camera-ready version accepted at NeurIPS 2026. Updated to the final conference format and included the NeurIPS checklist. 30 pages, 3 figure…

Vision-language-action (VLA) models can learn a wide range of robotic skills, but combining separately trained skills remains challenging. The original training data may be unavailable for joint retraining, while directly merging model weights can cause interference. We propose NestedVLA, which consolidates independently fine-tuned VLA policies without access to their original training data. Instead of merging all updates into one checkpoint, a hypernetwork embedded in the VLA generates action-expert weight updates for a selected task. Each task's compressed vision-language model update is stored separately, and both updates are added to a frozen base VLA to instantiate the corresponding policy. We evaluate NestedVLA by consolidating three policies fine-tuned with reinforcement learning on MetaWorld, ManiSkill, and CALVIN. It achieves 66.7% average success, improving on the strongest model-merging baseline by 41.1 percentage points. It performs within 1 percentage point of per-task fine-tuning on ManiSkill and substantially narrows the gap on CALVIN. In this three-policy setting, total storage is 1.5x that of a single full checkpoint. Code is available at https://github.com/WanyuGroup/NeurIPS2026-NestedVLA.

0.65

Keep Pace to Learn: Movable-Antenna Compensation of Channel Aging in Over-the-Air Split Learning

eess.SP cs.IT ·2026-10-16 ·Xianzhong Li, Yuzhi Yang, Tianhao Guo, Zhaoyang Zhang
highly cited
Abstract · v2: revised/extended; new title (v1: "Keep Pace to Learn: Movable Antennas Restore the Backpropagated Gradient Under User Mobility"); added …

Split learning (SL) trains a neural network across devices by exchanging intermediate activations and gradients. In over-the-air SL, the wireless channel serves as an inter-layer linear map, and when the forward and backward channels match, reciprocity and conjugation supply the backward product without channel estimation at the base station (BS). Passive user motion, which the learning link does not control, breaks this match. Movable antennas let the antenna move relative to the user device, and we use this freedom to compensate the displacement. For a stationary BS and a propagation environment quasi-static within each exchange, returning the antenna to its forward-transmission position restores the channel and yields a conditionally unbiased gradient estimate. We also derive when discrete positioning or mechanical movement can achieve this in time. A residual displacement below a quarter wavelength keeps the real part of the channel correlation positive for any angular spectrum. At zero effective offset, independent Gaussian jitter attenuates this correlation without reversing its sign, whereas a persistent offset can reverse it, and in arrays the tolerance depends on the receive-combining direction. These effects enter a nonconvex convergence bound, and a local analysis shows that partial mechanical compensation saves cycle time to first order at only a second-order loss in correlation. In CIFAR-10 experiments, a user terminal without channel state information stays at chance level under uncompensated walking motion, while discrete positioning stays within about 0.3 percentage points of exact restoration, including at a speed beyond the modeled actuator's limit.

0.58

Wait or Rotate: Statistical Repair of the Backward Pass in Over-the-Air Split Learning Under Channel Aging

eess.SP cs.IT ·2026-10-16 ·Xianzhong Li, Yuzhi Yang, Tianhao Guo
highly cited
Abstract · v2: substantially revised and retitled (v1: "Coherent but Discordant: Rotate, Don't Invert, to Backpropagate Through an Aging Channel"). Add…

Over-the-air split learning returns the receiver's loss gradient through the reciprocal channel, so terminal motion between the forward and backward transmissions corrupts the backward pass. We repair the aged return at a fixed receive array using only second-order statistics. Two numbers taken from the forward-backward cross-covariance, the alignment reachable by the best common phase and by the best unitary correction, bound the main term of every repair. In the 3GPP tapped-delay-line channels TDL-A/B/C, where every antenna ages by the same scalar, the two numbers coincide, and at a correlation zero all tested fixed corrections stay between 58% and 62% accuracy, whereas a return delayed to the next correlation lobe reaches 98.4%. In 3GPP clustered-delay-line channels a statistical unitary correction stays within 0.02 in mean cross-entropy of ideal channel knowledge in all six tested profile and speed combinations, while common-phase correction falls to 69% for CDL-A at 3 km/h and the tested delay loses up to 10 points. In a two-cluster model spatial correction is decisive only where the common-phase alignment nearly vanishes (64% to 95%). With a ResNet-18 on CIFAR-10, only the delayed return trains at a separable correlation zero (86% against at most 16%), and at 30 km/h in a clustered-delay-line channel the unitary correction reaches 80% where the common phase reaches 60%. Given geometry, displacement and angular statistics, a port permutation chosen by linear assignment reaches 99.5-99.7% accuracy for hexagonal and square arrays in oblique motion without pilots or digital feedback during training, where nearest-position matching drops to 89%.

0.57

HexaShell: A Universal Harness for Robot Coding Agents in Simulation and the Real World

cs.RO ·2026-10-16 ·Hongcheng Gao, Jianshu Zeng, Alex Zhou, Shijia Ge, Xuan Shangguan +11 more
benchmarkcode/datacommunity pick
Abstract · Code: https://github.com/HexaFuture/HexaShell

Agents built on vision language models (VLMs) now control robots by writing code, but the strongest of them refine task-specific code over many rounds of trial and revision. Such code tends to overfit the benchmark on which it was tuned, can exploit the benchmark's shortcuts, and is costly to carry over to a new benchmark or robot. We present HexaShell, a thin, task-agnostic harness designed to test how far general tools can go instead. We design a simple and complete set of 14 basic tools, 4 for motion, 5 for perception and 5 for state, which the agent calls in one Python cell per turn. Perception pairs SAM 3, which segments an object from a few words or a click, with NumPy and OpenCV geometry that turns its outlines in several cameras, or a depth image, into its position, size and grasp angle. The API, agent loop and sandbox are shared across all tasks; a new simulator or robot needs only an adapter of 350 to 1,450 lines and a prompt describing its world, and the only task-specific input is an optional skill of about a page of guidance. The same code runs on four simulated benchmarks, built on three simulators with two kinds of arm, and on real robots. With GPT-6-Astra and no task-specific input, HexaShell reaches success rates of 69.9% on RoboDojo, 59.3% on RoboCasa365, 78.3% on LIBERO-PRO and 82.2% on RoboTwin 2.0. With task skills, it reaches 85.0% (score 88.8), 85.6%, 96.7% and 96.8%, respectively, above the best published result we are aware of on each benchmark. The RoboCasa365 result is listed first on its leaderboard. On real dual-arm robots, HexaShell succeeds in 76 of 90 episodes across nine tasks.

0.56

Wait to Learn: Timing the Gradient Return in Over-the-Air Split Learning over an Aging Single-Antenna Channel

eess.SP cs.IT ·2026-10-16 ·Xianzhong Li, Yuzhi Yang, Tianhao Guo
Abstract · revised: theory reframed as instantiation of classical results, added line-of-sight extension

In over-the-air split learning over an aging single-antenna Rayleigh channel, the mean returned gradient equals the antenna-domain gradient scaled by the temporal correlation. Uncorrected training collapses in negative-correlation lobes. Near a correlation zero, no nonzero scalar calibration improves the ratio of mean power to innovation power for an individual transmission, and, on the tested grid, using sixteen times as many pilots recovers only 7 percent of the training gap. Returning at the first correlation extremum with phase correction, 1.3 ms later at 3.5 GHz and 15 m/s, achieves 90.2 percent accuracy, compared with 91.8 percent with CSI. In the tested line-of-sight setting, the near-zero gap becomes negligible.

0.56

Structured Perception Representations for Off-road Dynamics Learning

cs.RO ·2026-10-16 ·Micah Nye, Samuel Triest, Matthew Sivaprakasam, Isaiah Adu, Sebastian Scherer +1 more
published
Abstract · Accepted at CoRL 2026. 24 pages, 14 figures

Autonomously navigating off-road terrain at high speed demands accurate control, which often relies on dynamics models that capture complex interactions between the robot and terrain. In off-road regimes, terrain interaction is a major component of the vehicle's dynamics. Prior work attempts to capture this effect by training learned models of the vehicle's dynamics, conditioned on its perception of the local environment. However, said conditioning is often overly simplistic, ignoring the spatial structure of the dynamics problem. Our approach, SpatialBEV, exploits this spatial structure. At each step, we query the bird's-eye-view (BEV) features around the predicted state, conditioning the next prediction on current geometric and semantic cues. We additionally present a benchmarking approach that mines a dataset's most difficult scenarios. On datasets collected across seasons, SpatialBEV reduces prediction error by 17.1%, compared to physics-based and learned baselines. We further evaluate on hardware on high-speed ($\geq$10 m/s) and geometrically challenging courses with a 25.3% reduction in maximum cross-track error. More details can be found at https://theairlab.org/spatial-bev-dynamics .

0.55

Lifting Safety Through Reduced-Order Model Architectures: An Invariant Manifold Approach

eess.SY math.OC ·2026-10-16 ·Massimiliano de Sa, Aaron D. Ames
Abstract · 8 pages, 2 figures, Submitted to the American Control Conference (ACC) 2027

Reduced-order model architectures are often used to decompose complex safety-critical control design tasks into tractable subproblems. In this paradigm, one must ask: when does the safety of a reduced-order model lift to safety of the full system? Taking an invariant manifold perspective, we provide necessary and sufficient conditions for the existence of smooth, regular lifts of compact safe sets from reduced-order models. Our characterization relies only on asymptotic stability of an embedding of a reduced-order safe set, without additional robustness or rate conditions. We apply these results to study control barrier functions in reduced-order model architectures and lifts of safe sets through cascades. Further, we use our conditions to obtain topological obstructions to the tracking of safe, kinematic reduced-order models for mechanical systems. We conclude by using these obstructions to inform safe hierarchical control design for an underactuated satellite on SE(3).

0.50

Blind Precise Assembly: Proprioceptive Sim-to-Real Peg-in-Hole for Diverse Parts

cs.RO ·2026-10-16 ·Antonia Bronars, Younghyo Park, Kushal Kedia, Pulkit Agrawal
code/data
Abstract · 21 pages, 15 figures, 7 tables. Antonia Bronars and Younghyo Park contributed equally. Project page: https://younghyopark.me/blind-proprioce…

Every robot already carries high-resolution contact sensors: its own joints. Legged robots and humanoids exploit them to walk, run, and track whole-body motions without vision, but there the contacts are large and forceful. Is joint sensing also enough for precise, contact-rich assembly, where the robot must find a millimeter-scale opening, mesh gear teeth, or catch a thread through light contact? We study this question by training blind assembly policies that observe only a short history of joint positions, velocities, and their own actions, without visual, force, torque, or tactile sensing. The poses of both the fixture and the part are unknown to the policy, and the grasp is randomized substantially. The policy must therefore assemble the parts through contact, sensed exclusively through its own proprioception. Trained with reinforcement learning entirely in simulation over randomized parts, grasps, and contacts, such policies learn peg insertion, gear meshing, and nut threading. The peg insertion policy transfers zero-shot to the real world and inserts pegs of different shapes and sizes in 73% of trials. Controlled comparisons with matched visual policies in simulation show that blind policies generalize far better to unseen part shapes and sizes, though they can be slower. Training such policies needs thousands of different assemblies, each simulated with its clearance intact, so we also release an open-source generator that builds them procedurally from exact convex pieces for MuJoCo; it is more accurate and faster than convex decomposition or MuJoCo's signed-distance collision. Videos and interactive simulations are at https://younghyopark.me/blind-proprioceptive-assembly/.

0.49

Exploiting the Flexibility of LLM Inference for Frequency Regulation Service

eess.SY ·2026-10-16 ·Yimeng Sun, Zhaohao Ding
top venue
Abstract

While the widespread deployment of Large Language Models (LLMs) has significantly driven up power demands in computing clusters, the unique execution characteristics of LLM inference offer a novel opportunity to utilize these clusters as controllable loads. In this paper, we investigate the potential of LLM inference cluster to provide frequency regulation services through inference request scheduling. First, we model the problem as a hierarchical Markov decision process that includes request-to-replica dispatching and token-level resource allocation. Then, we introduce a reward redistribution network to address the temporal asynchrony between millisecond-level scheduling decisions and second-level frequency regulation signals. Furthermore, we develop an optimization framework based on a mean-field approximation to enable coordination among a large number of replicas under high request concurrency. Simulations using real-world data demonstrate that the proposed approach accurately tracks frequency regulation signals without compromising service-level objectives.

0.47

InstAgent: A Recursively Self-Improving Agent Operating System for Heterogeneous Robot Collaboration

cs.RO ·2026-10-16 ·Haoyang Luo, Junhao Qin, Jingyan Cao, Yi Huang, Guangjie Yang +5 more
Abstract · 9 pages, 3 figures

Heterogeneous robot teams enable complex, long-horizon tasks by combining complementary sensing, mobility, and manipulation capabilities across multiple embodiments. Yet existing systems struggle to retain execution-time corrections as reusable experience, maintain task-relevant context memory across space and time for reliable handoffs, and explicitly represent the inter-robot dependencies required for structured collaboration. In order to address these challenges, we present InstAgent, a recursively self-improving (RSI) Agent Operating System for heterogeneous robot collaboration. InstAgent organizes team behaviors as agentic behavior trees (BTs) that explicitly model capability assignment, inter-robot dependencies and handoff checks. The RSI mechanism updates these collaboration programs from execution feedback and preserves successful revisions for reuse in future tasks, while shared multimodal spatio-temporal memory provides context for planning and handoff validation. Experiments on 20 real-world multi-robot tasks with three high-level models show consistent gains in task success, first-attempt completion, and execution efficiency.

0.46

Causal adaptive intensity estimation for non-stationary point processes: a Kalman-kernel duality framework

eess.SP stat.ME ·2026-10-16 ·Hanwu Qin, Dianhai Wang, Meng Zhang, Huangjun You, Zhengyi Cai
Abstract

Estimating the instantaneous conditional intensity of a non-stationary point process from a single realization of discrete event timestamps is a fundamental problem in computational neuroscience, quantitative finance, and network monitoring. Traditional adaptive kernel methods achieve variable bandwidths through Abramson's square-root law, yet existing implementations require non-causal access to the complete event record and incur $\mathcal{O}(N^2)$ or at best $\mathcal{O}(N \log N)$ computational cost. Causal alternatives based on reciprocal inter-arrival times, meanwhile, suffer from systematic divergence under Poisson and bursty arrival dynamics due to Jensen's inequality. This paper addresses both limitations by establishing a formal duality between steady-state Kalman filtering and adaptive exponential kernel smoothing. An endogenous state-space observer is constructed in which Poisson counting statistics dictate the measurement noise ($R \propto λ$) and geometric diffusion governs the process volatility ($Q \propto λ^2$). Under these physically motivated axioms, the algebraic Riccati equation autonomously yields an effective smoothing bandwidth scaling as $τ\propto λ^{-1/2}$, giving a causal, recursive, $\mathcal{O}(1)$ per-step realization of square-root adaptive bandwidth scaling. A second-order constant-velocity extension produces a $λ^{-1/4}$ scaling that reduces approximation bias, and a probabilistic dormant-state gating mechanism suppresses phantom intensity leakage during structural zeros. Synthetic benchmarks together with neural and financial data show that the observer adapts across orders of magnitude in intensity with an empirical per-step cost of approximately 30\,$μ$s.

0.43

Expand-and-Prune: Test-Time Candidate-Space Adaptation for End-to-End Autonomous Driving

cs.RO cs.CV ·2026-10-16 ·Fabian Schmidt, Karol Fedurko, Markus Enzweiler, Abhinav Valada
publishedcode/data
Abstract · Accepted to IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2027

Candidate-based end-to-end driving planners reason over a finite set of trajectory hypotheses. However, for sparse planners, a limited candidate set may provide insufficient support for scene-relevant maneuvers, while fixed downstream candidate budgets do not account for prediction uncertainty. To address these challenges, we introduce Expand-and-Prune, a test-time framework that jointly adapts candidate support and capacity without updating learned parameters. The framework expands the native candidate set using semantic driving intent and temporally propagated predictions and then uses uncertainty in intermediate trajectory scores to adaptively prune the resulting candidates. Across four heterogeneous candidate-based planners and four benchmarks, Expand-and-Prune consistently improves planning performance. It achieves state-of-the-art results on LangAuto, increasing the Driving Score from 68.2 to 75.9, while providing further improvement on Bench2Drive, Fail2Drive, and NAVSIMv2, with particularly strong improvements in out-of-distribution scenarios. Moreover, in its full instantiation, 72.5% of selected trajectories originate from the newly introduced intent-guided or temporal candidates, confirming that candidate-space expansion provides useful hypotheses beyond those generated natively by the planner. The code is available at https://github.com/iis-esslingen/Expand-and-Prune.

0.41

Bio-Inspired Magnetic Navigation Using Gradient Descent With Rao-Blackwellized Particle Filtering

cs.RO ·2026-10-16 ·Zayed Herma, Andrea Vassilev, Philippe Cardin, Jean-Michel Leger
published
Abstract

Navigation in GPS-denied environments is commonly based on magnetic maps, which are used to correct trajectory drift and enable relocalization. However, such navigation systems suffer from a strong dependence on magnetic maps that are complex, heavy, and costly to construct. Bio-inspired magnetic navigation (BIMN) does not rely on magnetic maps, making it simpler and more frugal. In this paper, we present a gradient descent BIMN method based on the Rao-Blackwellized particle filter (RBPF) for local gradient estimation. We show that magnetic navigation can be performed without magnetic maps, as gradient descent naturally attracts the system toward the destination, resulting in a simple and memory-efficient approach.

0.41

Net-Zero Power UAVs: Facilitating Quantum HRLLC in Multi-Hop UAV Repeaters

eess.SY ·2026-10-16 ·Ali Ranjha, Georges Kaddoum, Halim Yanikomeroglu
publishedbenchmarkhot direction
Abstract · 5 pages. Published in IEEE Wireless Communications Letters. This work was supported by the Postdoctoral Research Fellowship from the Fonds d…

In this paper, we propose a novel joint optimization framework for autonomous, quantum-enabled UAV repeater networks designed to support hyper-reliable low-latency communications (HRLLC) in challenging environments such as post-explosion chemical plant operations. The proposed system uses a symmetric dual-UAV architecture with three sequential quantum channels operating under strict energy harvesting (EH) constraints. We formulate a highly challenging non-convex mixed-integer non-linear programming (MINLP) problem. This model jointly determines qubit allocation, 2-D uncrewed aerial vehicle (UAV) spatial positioning (distance and altitude), and transmit powers to minimize the total decoding error. Crucially, our framework explicitly models real-world quantum physics, including decoherence, gate imperfections, and entanglement degradation. Since the objective function is demonstrably non-convex, to efficiently solve the MINLP, we propose a quantum-resource-optimized GA (QRO-GA). The results of our numerical simulations show that the QRO-GA yields better results than benchmark methods and it is highly efficient, with the convergence speed significantly increasing with the total qubit allocation. Convergence is achieved in as few as 5 iterations for higher qubit allocation. Moreover, the results reveal a critical U-shaped tradeoff, proving that an optimal latency distribution within a strict $80\ μ$s transmission window is strictly required to maintain HRLLC reliability. In addition, we find that the QRO-GA successfully optimizes the 2-D UAV positioning to minimize total decoding error. Finally, it balances consumed and harvested power to achieve net-zero power UAVs system.

0.40

SCOPE: Selective Certification of Program Edits for Embodied Heuristic Learning

cs.RO ·2026-10-16 ·Junyang Shu, Bingqing Wei, Jishuo Li, Zheqi He, Yongtao Wang
code/data
Abstract

Embodied heuristic learning must balance strategic adaptation with the preservation of previously successful behavior. Local repairs may leave structural failures unresolved, whereas broader revisions can introduce regressions that are costly to detect through repeated environment interaction. We introduce SCOPE (Selective Certification Of Program Edits), which uses failed and successful executions as complementary evidence for program revision. Frontier-guided program search annotates a shared diagnostic graph with execution evidence to distinguish supported progress from unresolved checks. Selective behavioral certification uses stateful API replay to restore recorded responses and mutations, checking whether edits preserve selected historical successes. Diagnosis and certification feedback, supplemented by sparse environment diagnostics, guide consecutive repairs across configurations. SCOPE outperforms the strongest baseline in both simulation and real-world tasks. On LIBERO-PRO and Robosuite, it achieves mean success rates of 80.66% and 84.48%, with respective gains of 6.51 and 3.48 percentage points. Our code is available at https://github.com/VDIGPKU/SCOPE.

0.38

LY-GWM-RoboCasa-Human300: Prediction-Guided Action Selection with a Graph Dynamics Model

cs.RO ·2026-10-16 ·TianLiang Ma, Ting Ma, ZhongYuan Ma
benchmarkcode/data
Abstract · 15 pages. Technical report. Code and evaluation materials: https://github.com/SEIN-LYGWM/LY-GWM-RoboCasa-Human300-eval

We present LY-GWM-RoboCasa-Human300, a policy that combines a Human300-fine-tuned GR00T-N1.5 action generator, an action-conditioned graph dynamics model, and a learned binary state-reward scorer. At inference, the policy samples four action chunks, predicts the future latent features and robot state for each, and executes the candidate with the highest predicted-state reward logit. Predictions therefore influence the actions sent to the environment. The evaluated system uses a 16-step prediction horizon and keeps all three components frozen during evaluation. In the local evaluation on 50 RoboCasa 1.0.1 tasks in the pretrain kitchen split, with 50 episodes per task, the system records 207/2,500 successes (8.28%), compared with 181/2,500 (7.24%) in the historical self-trained GR00T-N1.5 baseline. The observed difference is 26 successes, or 1.04 percentage points. The atomic group improves from 141/900 to 150/900. Separately, the RoboCasa maintainers re-evaluated the submission and reported 354/1,500 successes (23.6%) across 50 tasks, with leaderboard rates of 48.3%, 16.3%, and 3.1% for Atomic-Seen, Composite-Seen, and Composite-Unseen. The local and official results are reported separately; their difference is not attributed to a model improvement. Recorded local audits identify 212,700 changed environment-input chunks, including 28,223 in atomic tasks. We describe the actual training stages, implemented inference mechanism, evaluation provenance, and public reproduction entrypoints. These results document an active prediction-and-selection system and a positive historical difference.

0.37

What Makes a Good Symbol for Hybrid Discrete-Continuous Planning?

cs.RO ·2026-10-16 ·Yiyuan Pan, Changliu Liu, Jiayuan Mao
Abstract

The physical world has continuous state and action spaces, yet robot planning often benefits from hybrid discrete-continuous formulations that explicitly expose combinatorial structure, such as contact modes or discrete primitive actions. Existing methods typically rely on human-defined symbol sets or symbols derived from language supervision, without asking what makes a symbol computationally useful. In this paper, we take an inference-algorithm--centric view to define usefulness: a good symbol should expose the discrete commitments that simplify solving the original planning problem. We first introduce the concept of solver-sufficient symbolization by studying finite-horizon piecewise-affine systems, for which solver sufficiency can be precisely characterized, and develop a counterexample-guided algorithm for symbol invention. We then extend this principle to invent symbols for optimization objectives parameterized by general neural networks as energy functions, showing that solver-sufficient symbols can nevertheless be induced from their continuous optimization landscapes. Experiments on simulated and real-robot tasks show that the resulting symbols reduce solving cost while preserving solution quality, while also improving interpretability and steerability. Furthermore, our theoretical and algorithmic insights help explain why certain human-defined symbols are computationally useful.

0.37

TransNav: Transformer-Based End-to-End Navigation Method for Static and Dynamic Quadrotor Obstacle Avoidance

cs.RO ·2026-10-16 ·Jianghan Zhu, Zhuo Li, Yanjun Zhang, Jian Sun
top venue
Abstract · 8 pages, 6 figures, 2 tables. Published in IEEE Robotics and Automation Letters (RA-L)

Obstacle avoidance for autonomous quadrotors in cluttered environments remains challenging, especially when both dense static obstacles and moving obstacles are present. Existing methods are often tailored to either static or dynamic settings, which always limit their adaptability across diverse environments. This paper presents TransNav, a transformer-based end-to-end navigation method for quadrotor obstacle avoidance in both static and dynamic environments. TransNav employs a patch-based point-cloud encoding scheme that combines local feature extraction with global contextual modeling, enabling compact yet informative environment representation. The method integrates perception and navigation in a single architecture and directly predicts a finite-horizon action sequence, from which the first action is executed. Moreover, we develop an unsupervised learning strategy by incorporating a lightweight point-mass dynamics model, which directly optimizes trajectory-level costs over the predicted sequence through numerical gradient propagation. Thus, the policy can be efficiently trained without expert demonstrations. Simulation and real-world experiments show that the proposed method achieves strong performance in dense static environments and competitive obstacle-avoidance capability in dynamic scenarios, while maintaining low inference cost for onboard deployment.

0.36

Pedicel-Targeted Strawberry Harvesting via Imitation Learning on a Low-Cost Teleoperated Robotic Arm

cs.RO ·2026-10-16 ·Nitin Rai, Won Suk Lee, Hongyoung Leon, Heping Zhu
publishedcode/data
Abstract · 4 pages, 3 figures. This paper was peer-reviewed and accepted in the IEEE/RSJ International Conference on Intelligent Robots and Systems (IR…

Precise pedicel-targeted grasping in strawberry harvesting remains a critical bottleneck due to the pedicel's small diameter (1.3-3 mm), color similarity with surrounding foliage, and precise force control required for clean abscission zone detachment. As a market-grade crop, strawberry aesthetic value makes bruise-free pedicel detachment essential over direct fruit grasping. Classical approaches struggle to unify vision models, motion planning, and contact manipulation into a robust framework, particularly in occluded canopy environments. Therefore, this work presents a visuomotor imitation learning pipeline for pedicel-targeted strawberry harvesting that unifies perception, motion planning, and contact manipulation from collecting human teleoperation demonstrations, training an Action Chunking with Transformers (ACT) policy with a ResNet-50 visual backbone, and deploying the learned policy autonomously on a Jetson AGX Orin edge platform at 15 Hz. The trained policy achieved consistent loss convergence from 2.84 to 0.091 over 100K gradient steps on 130 demonstrations. We have also established a reproducible open-source pipeline with publicly released dataset and model weights for the agricultural robotics community. Full dataset is available at, https://huggingface.co/datasets/nitindominicrai/strawberry_pedicel_grasp_teleoperation and trained policy at, https://huggingface.co/nitindominicrai/act_strawberry_pedicel. To replicate this work, all code, configurations, and deployment instructions are publicly available at, https://github.com/nitin-dominic/autonomous_strawberry_pedicel_harvesting.

0.36

MambaAct: Efficient Mamba Action Experts for Flow-Matching Vision-Language-Action Models

cs.RO cs.CV ·2026-10-16 ·Khuyen Pham, Tran Nguyen Le, Edoardo Fazzari, Cesare Stefanini
code/data
Abstract

Vision-language-action (VLA) models have shown strong promise for generalist robotic manipulation, but their inference cost remains a major bottleneck for real-time control. This limitation is especially pronounced in flow-matching VLAs such as $π_{0.5}$, where the action expert is evaluated repeatedly during iterative denoising. We introduce MambaAct, an efficient Mamba action expert for flow-matching VLA policies. MambaAct retains the pretrained visual-language prefix encoder of $π_{0.5}$, while replacing its transformer-based action expert with a compact Mamba-based action expert. Within this expert, Mamba serves as the action mixer that models dependencies across the noisy continuous-action sequence. To preserve visual-language conditioning without repeated token-level cross-attention, MambaAct uses a lightweight Context Compressor to map Vision-language model (VLM) prefix tokens into a compact conditioning signal that gates and modulates each Mamba action block. MambaAct uses a $10\times$ smaller action expert and reduces flow-matching denoising latency by $64.8\%$ on LIBERO and $58.8\%$ in real-world deployment, while improving mean LIBERO success by $1.08$ percentage points across three independent fine-tuning seeds and maintaining gains on LIBERO-Plus and real-world benchmarks. Our code and model weights are publicly available at https://github.com/giakhuyendihoc/MambaAct.

0.36

CLARO: Clutter Aware Robot Orchestrator for Open-Vocabulary Grasping using VLM Agents

cs.RO cs.CV ·2026-10-16 ·Sungsu Kim, Vineet Bhat, Prashanth Krishnamurthy, Ramesh Karri, Farshad Khorrami
published
Abstract · Accepted to WACV 2027

Robots that follow natural-language instructions unlock a wide range of autonomous applications in households, warehouses, and labs. Real-world environments are diverse and often heavily cluttered, requiring robots to plan over multiple steps, act, and replan as they work toward a goal. We present CLARO (CLutter-Aware Robot Orchestrator), a VLM agent that orchestrates a set of perception and manipulation skills: open-vocabulary object detection, segmentation, 6-DoF grasping, and non-prehensile pushing to locate and retrieve a language-specified target in clutter. CLARO maintains a memory of its past actions to support long-horizon execution, and emits manipulation targets (grasp poses/3D waypoints) allowing transfer across robot arms without task-specific training. We evaluate CLARO against a range of methods on a dynamic simulation environment and standard image-based benchmarks, demonstrating strong generalization across object classes, lighting, and clutter levels. CLARO attains an average success rate of 89.1%, a relative improvement of over 16% on the strongest competing method in high cluttered scenes. We further analyze performance along grounding and grasping axes to characterize where each method succeeds and fails. Finally, we deploy CLARO on a robot arm across diverse cluttered scenes, achieving a success rate of 88.8%, and demonstrating strong sim2real generalizability.

0.34

Hybrid Cones: Set Representation with Sharp Unions of Minimal Overhead for Nonlinear and Switched Linear Systems

eess.SY ·2026-10-16 ·Daniel Silvestre
Abstract · 19 pages

Reachability analysis and set-based estimation for nonlinear systems require set representations capable of capturing nonconvex geometry. Hybrid Zonotopes (HZs) have recently gained attention because they can encode a union of up to $2^n$ polytopic components. Recent work has further shown that several HZ operations can be made \textit{sharp}, meaning that relaxing the binary variables yields exactly the convex hull of the represented set. However, no known HZ construction provides a sharp union of two convex sets that adds only one binary generator and no further continuous generators or equality constraints to those of the operands. This limitation affects scalability in repeated union operations, ReLU neural-network reachability, and nonlinear propagation methods that cover graph sets by unions of simpler sets. This paper identifies a connection between the exact closed-form convex-hull operation for Constrained Convex Generators (CCGs) and the binary-generator structure underlying HZs. We propose Hybrid Cones that can encode up to $2^n$ norm-cone components. Hybrid Cones inherit the logical structure of HZs while allowing richer convex components, including polytopic and norm-based sets. They support sharp unions requiring only one added binary variable and can represent unions and intersections of polytopes and ellipsoids. As a key application, we derive an exact representation of the ReLU graph using only two continuous and one binary generators. Numerical experiments for a localization problem with a nonlinear energy measurement function approximated by a neural network and the stability verification of the closed loop of a double integrator corrupted by a nonlinear disturbance with a saturated Linear Quadratic Regulator (LQR) controller show substantial reduction in representation size and reduced conservatism in comparison with HZs.

0.34

Switch-Time-Aware Hybrid Model Predictive Control Using the Hybrid Minimum Principle

cs.RO eess.SY ·2026-10-16 ·Valentin Gaucher, Wenlong Zhang
Abstract · 8 pages; 4 figures, 2 tables

This article presents a time-aware hybrid model predictive control (MPC) framework that simultaneously optimizes the objective with respect to both continuous control inputs and event switching times. Rather than relying on a fixed time grid, the switching instant is lifted into the decision vector and treated as an explicit optimization variable. The resulting transcription is a smooth nonlinear program whose Karush-Kuhn-Tucker (KKT) conditions are consistent with the Hybrid Minimum Principle (HMP) event-time stationarity condition, so first-order hybrid sensitivities are recovered implicitly through the KKT system rather than by evaluating saltation matrices. Each MPC step solves a single sparse Quadratic Program (QP), jointly optimizing control inputs and impact timing in real time. On a bouncing mass landing on an oscillating platform, the method attaches in half the time of fixed-grid saltation MPC, reduces cumulative control effort by 30%, and removes the post-impact rebounds, at a 15% increase in per-cycle solve time. It matches the cumulative effort of offline time-aware solvers to within 4% while running at 2.4 ms per cycle against 281 ms for a complementarity-based formulation of the same problem, of which a 10.6x reduction is attributable to the transcription alone at a fixed solver.

0.34

Towards Reinforcement Learning for Space Robotics: Fast Training and Reliable Real-World Transfer

cs.RO cs.AI ·2026-10-16 ·Abhishek Naik, Michael Wu, Michael O'Sullivan, Colin Bellinger, Yunli Wang
published
Abstract · Presented at the Space Robotics Workshop at IROS 2026

This work demonstrates the use of a fast simulation-based training framework to learn control policies using reinforcement learning. The hardware requirements are minimal and the simulation-learned policies transfer reliably to the real world. With this setup, we show that a quadruped robot can learn to traverse diverse real-world terrains reliably---including in a lunar-analog facility---after less than 20 minutes of training on a single consumer-grade GPU. These results provide a glimpse of a future where reinforcement learning is a crucial tool in developing more capable and robust space robotic systems. Video demonstrations are available at https://youtu.be/vUFJiZdJaQ0.

0.33

From Sliding Windows to Self-evolving Lifelong Memory: A Dual-perspective Survey of Memory Paradigms for Long-Horizon Robotics VLA Models

cs.RO ·2026-10-16 ·Yong Wang, Boyu Shen, Xinwei Zhang, Yizheng Li, Keyuan Luo
surveycode/data
Abstract · 10 pages, 1 figure

Long-horizon robotic tasks expose a fundamental limitation of current Vision-Language-Action (VLA) models because their restricted temporal context pre-vents effective accumulation and reuse of past experiences. As a result, memory has become a fundamental component for enabling embodied agents to perform long-horizon reasoning and adaptive decision-making. This survey presents a systematic taxonomy of VLA memory mechanisms based on memory scope and persistence, categorizing existing approaches into five representative paradigms and one emerging frontier. Based on this taxonomy, we compare representative VLA memory systems in terms of memory characteristics, architectural com-plexity, and open-source maturity, revealing a clear trade-off in which increasing architectural complexity is generally accompanied by decreasing open-source ma-turity. Finally, we discuss the key challenges in building full-lifecycle memory systems, including efficient memory management and lifelong knowledge evolu-tion. Overall, this survey provides a unified perspective on memory architectures for VLA agents and offers insights into the development of lifelong embodied in-telligence.

0.31

Continuous-Phase Frequency-Hopping MIMO Signaling and Processing for ISAC

eess.SP ·2026-10-16 ·Yi Wang, Xianxiang Yu, Jing Yang, Tao Fan, Linlong Wu +2 more
Abstract · 22 pages, 13 figures, including appendices

Frequency-hopping MIMO integrated sensing and communications (FH-MIMO-ISAC) supports agile multidimensional signaling but faces challenges in phase-continuous transmission, demodulation without channel-phase information, and coherent processing of multi-frequency echoes. We propose a continuous-phase frequency-hopping (CP-FH) MIMO ISAC waveform with matched communication and radar processing. Information is conveyed jointly by frequency combinations and continuous-phase-modulation symbols. CPM phase accumulation and frequency-hopping linear-frequency-modulation phase compensation maintain phase continuity and reduce spectral leakage. We derive exact inter-element orthogonality conditions, residual cross-correlation bounds, and a bound on averaged transmit-beampattern ripple. For communication, projection-energy demodulation requires no channel-phase information and approximates least-squares detection under weak inter-branch correlation. It separates frequency-combination search from CPM-symbol decisions, reducing complexity while retaining direct joint-search decisions. For radar, structure-matched coherent processing combines space-time-frequency matched filtering, velocity-consistent Doppler focusing, and frequency-aperture focusing to form a three-dimensional coarse-range--velocity--fine-range response. It reduces Doppler-mismatch loss and extra detections from periodic range grating lobes in conventional pulse-Doppler processing. We further derive the joint range response, output signal and noise powers, and receive beampattern. The analysis characterizes target-discrimination limits under periodic ambiguity and shows that the proposed processor preserves frequency-aperture range resolution and full-aperture angular resolution, with equal output noise power under the stated conditions.

0.31

IMU-Aided Motion Compensation of RSS Measurements for Dynamic Visible Light-Inertial Localization

eess.SP ·2026-10-16 ·Xiao Sun, Zhenqi Zheng, Jianzhu Huai, Xingxing Li, Wanwei Zhang +1 more
code/data
Abstract

Visible Light Positioning (VLP) offers high-precision indoor positioning using existing LED infrastructure. However, in VLP systems using Photodiodes (PDs) as receivers, rapid movement of the PD causes measurement inaccuracies as the Discrete Fourier Transform (DFT) processing yields averaged rather than instantaneous Received Signal Strength (RSS) values. To address this challenge, we propose an RSS correction algorithm that integrates Inertial Measurement Unit (IMU) data with VLP observations to compensate for motion-induced distortions. Our approach leverages IMU-derived motion parameters to model the temporal variation of optical signals during the DFT window, enabling instantaneous RSS estimation. The algorithm is embedded within both tightly and simply coupled VLP/INS integration frameworks that estimate the pose of the moving receiver. This work targets highly dynamic indoor navigation. Simulations validate the correction model, and total-station-tracked experiments confirm reduced RSS discrepancies against model references. Across eight held-out high-dynamic handheld trials with peak translational and angular speeds of 2.25 m/s and 284 deg/s, correction reduces the pooled 3D position and PD-normal inclination root mean square errors relative to the LiDAR/IMU reference with a shared IMU by 56.0% and 71.9%, respectively. Both datasets and source code are publicly available.

0.31

SenseSwarm: Swarm Exploration via Neural Structure Estimation with Shared World-Anticipation for Rapid Mapping

cs.RO ·2026-10-16 ·Haohua Que, Haojia Gao, Mingkai Liu, Hoiian Au, Qian Zhang +5 more
benchmark
Abstract · 9 pages, 7 figures

A robot swarm covers a building faster with more robots, but the map is only as complete as the space the robots have physically visited. We target the amount of an unknown building a decentralized swarm knows at each instant, ahead of physical coverage. SenseSwarm serves a lightweight learned map predictor on the swarm's shared occupancy map and folds the predicted structure back into a stigmergic guidance field. Its 709K-parameter predictor is fine-tuned on the swarm's own multi-front observations, which we show is necessary: a single-robot predictor collapses on such views. We build a benchmark of twelve real building floors from HouseExpo, HKUST, and KTH, up to 24,000 square meters, and run 1,620 simulations across nine coordination methods, swarm sizes 32 to 512, and three seeds. On physical coverage the swarm is near parity with a greedy frontier planner, as expected. On map reconstruction, every non-predictive baseline needs 1.3 to 1.9 times longer to reach an 80 percent known map, and SenseSwarm holds an equal-time area lead of about 670 square meters, up to 1,359. Folding the prediction into the shared field matters: the same predictor applied per robot matches our speed but suffers a 6.9 times higher inter-robot contact rate at 512 robots. The prediction is computed once per shared map, so its cost stayed flat from 32 to 512 robots.

0.30

Covert Digital Semantic Communication With Movable Antennas: Constellation Shaping, Spatial Optimization, and Semantic Mapping

eess.SP cs.IT ·2026-10-16 ·Yunfan Bai, Yuwen Qian, Hao Jiang, Yiyang Ni, Feng Shu +2 more
Abstract · 13 pages, 9 figures. Submitted to IEEE Journal on Selected Areas in Communications (JSAC)

Digital semantic communication (SemCom) is emerging as a bandwidth-efficient paradigm for next-generation wireless networks by delivering the meaning of a source over finite constellations. However, the broadcast nature of wireless channels exposes transmissions to adversarial wardens, while discrete symbols deviate from the Gaussian signaling assumed by conventional covert communication. In this paper, we propose a covert digital SemCom system assisted by movable antennas (MAs), where the constellation, the antenna positions, and the semantic transceiver are jointly designed to conceal the transmission while preserving the semantic fidelity. Specifically, the maximum covert rate against the optimal detector with unknown noise level is derived in closed-form, which reveals that the legitimate signal-to-noise ratio (SNR) scales with the moment-matching order of the constellation and the spatial gain provided by the MA array. A Gauss-Laguerre ring constellation is then designed to raise the moment-matching order under the peak power constraints, and a covert symbol mapping with rank-to-ring assignment and scrambling is developed to carry the semantics in the ordering of the latent magnitudes. Guided by the derived semantic distortion bound, the semantic encoder and decoder are trained end-to-end through an invertible flow over the designed constellation and MA array. Simulation results demonstrate that the proposed scheme supports a higher achievable covert rate and improves the reconstruction quality by 24% compared to state-of-the-art SemCom schemes.

0.30

Embodiment-Aware Planning for Heterogeneous Bimanual Manipulation

cs.RO ·2026-10-16 ·Sukhvansh Jain, Anurag Maurya, Shivam Vats, Narendra Kumar Dhar, Ravi Prakash
Abstract · Please find the project website here: https://sukhvansh2004.github.io/Bimanual-Manipulation/

Many manipulation tasks that are effortless for dexterous human hands remain challenging for robotic systems equipped with simple, task-specialized grippers. Industrial platforms commonly rely on end-effectors such as suction cups or parallel-jaw grippers, yet no single mode can reliably satisfy the diverse constraints encountered across different phases of complex manipulation tasks, including pickup, reorientation, insertion, and constrained placement. In this work, we investigate whether complementary simple grippers within a heterogeneous bimanual setup can collectively realize tasks that exceed the capability of any individual gripper alone. Given a single human demonstration specifying the desired object motion, we formulate embodiment-aware execution as a finite-horizon optimization over arm-trajectory assignments. At each segment, the planner determines whether single-arm manipulation or a coordinated dual arm handover is required, jointly reasoning over heterogeneous grasp feasibility, motion constraints, manipulability, and cross-gripper compatibility. Exploiting this formulation, we compute globally consistent assignments with explicit switching regularization. Experiments on a dual Franka FR3 platform equipped with suction and parallel-jaw grippers demonstrate improved feasibility and reliability in tasks that are infeasible under single-gripper execution.

0.29

PASS-VLN: Passability-Aware Local-Command Vision-and-Language Navigation

cs.RO ·2026-10-16 ·Cong Zou, Xiangsheng Shi, Xuanhui Wang, Fubin Wang, Guangjun Ge +2 more
benchmark
Abstract · 9 pages, 4 figures

Reliable vision-and-language navigation (VLN) requires not only understanding instructions but also determining whether and how a robot can physically execute them. However, existing VLN benchmarks typically assume executable instructions and focus on long-horizon tasks. This leaves body-dependent local execution, including passability and collision-free execution, largely unexamined. Evaluating these factors within long-horizon tasks would entangle local physical execution with task planning and long-term spatial memory, making navigation failures difficult to attribute. We therefore focus on grounding local navigation commands in executable motion under environmental and robot body constraints. We present PASS-VLN, a passability-aware local-command VLN benchmark and model. The benchmark holds the scene, instruction, start pose, and reference route fixed while varying robot body width, requiring target completion when the instructed passage is feasible and stopping before blockage otherwise. Our model encodes body width through obstacle inflation in a depth-derived local bird's-eye view, fuses geometry with visual and linguistic context, and predicts discrete latent action codes for continuous closed-loop execution. Experiments show that PASS-VLN improves collision-free success by approximately 12 percentage points over a strong VLN model fine-tuned on our benchmark, while adapting reach-versus-stop behavior to body constraints and generalizing to unseen body widths. When integrated with a high-level planner, PASS-VLN achieves higher success with fewer local commands, while real-world deployment further demonstrates its effectiveness as a low-level executor. The benchmark and model will be released.

0.28

VLA-Fabric: Learning Interaction for Composable Multi-Arm VLA Agents

cs.RO ·2026-10-16 ·Chenglong Zhang, Shuaijun Liu, Feiyang You, Shuyang Hao, Chengyu Wu +1 more
Abstract · 17 pages, including references and appendices. Project website: https://nebulis-lab.com/VLA-Fabric/

Vision-language-action (VLA) models are rapidly expanding what individual robots can accomplish, yet many real-world manipulation settings require multiple arms to operate as a coordinated system. Existing approaches organize multi-arm coordination in different ways, including centralized joint policies, high-level task or role assignment, and locally executed policies that react to teammates. However, how complete action-producing VLAs should explicitly interact with one another during action generation remains largely unexplored. We introduce VLA-Fabric, which organizes one complete VLA per arm while preserving local perception and action ownership, and enables coordination through learned interaction across policy boundaries. We first systematically study where such interaction should occur, how peer information should be aggregated, and how the interaction should be learned, revealing complementary coordination functions spanning shared context, peer-specific latent information, and action-stage interaction. We then re-instantiate these functions in a structurally different VLA architecture, showing that the coordination principle is not tied to a particular tensor layout or backbone implementation. Finally, we compose teams of three and four arms and evaluate them in simulation and physical experiments. On all four physical tasks, Full interaction improves success over independently trained local policies by 16--40 percentage points. Together, these results suggest that explicit interaction among complete VLA policies offers a scalable path toward networked embodied intelligence.

0.27

Sequence-to-Sequence Neural Surrogate Modeling of Transient Responses in High-Speed Links with Variable Design Parameters

eess.SP cs.CEcs.LG ·2026-10-16 ·Julian Withöft, Shayan Ghanbari, Werner John, Emre Ecik, Ralf Brüning +1 more
Abstract · 12 pages, 13 figures, 5 tables. This work has been submitted to the IEEE for possible publication. Copyright may be transferred without noti…

This paper presents sequence-to-sequence (seq2seq) neural surrogates for high-speed transient response modeling and signal integrity analysis in fixed and variable design parameter spaces. To identify the structural requirements for prediction accuracy, multiple neural architectures were evaluated with respect to different training strategies such as sample-based and unit interval-based, as well as input configurations such as TX-only and TX+RX. For fixed design parameters, feedforward neural networks achieve accuracy comparable to recurrent models while requiring lower computational cost, as the static mapping can be approximated by fully connected and non-sequential representations. Variable design parameters alter the electromagnetic characteristics of the interconnect, which creates parameter-dependent temporal dynamics that require sequence-aware architectures to resolve the evolving interactions within the transient response. The results based on a 14-dimensional variable design parameter scenario show that while autoregressive TX+RX feedback suffers from severe phase drift and error propagation under variable conditions, non-autoregressive TX-only inference leads to stability. To overcome the structural limitations of standard decoders, a recurrent Early Fusion LSTM is proposed, which injects design parameters at the initial encoding stage. This approach achieves superior prediction accuracy, with an RMSE of 31.0 mV and an R2 of 99.51% for transient waveform prediction, an RMSE of 30.7 mV and an R2 of 99.57% for eye-height prediction, and an RMSE of 19.5 ps and an R2 of 99.94% for eye-width prediction. These results demonstrate that neural seq2seq surrogates are accurate for transient response modeling and signal integrity analysis in complex variable design parameter spaces.

0.27

LAW: Unifying Language, Action, and World Modeling for Agentic Physical Intelligence

cs.RO ·2026-10-16 ·Jiajun Cao, Kai Tang, Bohua Zhang, Zezhong Qian, Ying Li +16 more
code/data
Abstract

World-action models jointly predict actions and future world states, but long-horizon manipulation further requires language reasoning to plan subtask objectives, assess execution progress, and guide action selection. We present LAW, a closed-loop VLM--WAM framework that unifies language reasoning, action generation, and world modeling through semantic subtask progress. Within the LAW agent, a hybrid-expert vision-language model plans the remaining subtasks, estimates the values of and fuses actions proposed by a world-action model, and asynchronously assesses execution status to support continued execution and subtask transitions. Within the world-action model, Mixture-of-Transformer Experts integrates modality-specific and shared computation for joint video--action modeling. Dual-stream hierarchical visual memory maintains temporal context for both models, while subtask-local value learning aligns action evaluation with semantic objectives and preserves supervision from successful prefixes of failed trajectories. LAW achieves 94.21% average success across 50 RoboTwin tasks. Replacing the WAM backbone with OpenWAM-$alpha$ further raises the average success rate to 95.56%, establishing a new open-source state of the art on RoboTwin. Despite having only 2B parameters, our VLM also outperforms variants using Embodied-R1.5 and GPT-5.5 High. On three long-horizon real-robot tasks, LAW achieves 65% average success, compared with 28.33% for $pi_{0.5}$ and 25% for Fast-WAM.

0.27

BeJEPA: A Goal-Conditioned Humanoid Behavior World Model via Unsupervised RL

cs.RO ·2026-10-16 ·Liyun Yan, Haoyang Li, Yang Zhang, Jialing Zhu, Zhanxiang Cao +9 more
code/data
Abstract

Humanoid control requires coordinating whole-body posture with position and heading in the world frame. Mainstream controllers imitate trajectories in the root frame, but struggle to regain the target world pose after displacement. We introduce BeJEPA, a goal-conditioned Behavior World Model learned through unsupervised reinforcement learning from human motion priors and robot interaction. BeJEPA learns a shared representation of body configuration and world placement through a joint-embedding predictive architecture (JEPA). A Dynamic Predictor predicts the next state embedding under an action, while a Successor Predictor summarizes a goal-conditioned policy's expected discounted future state embeddings in a single vector. During joint training, Dynamic grounds the initial representation in observed physical transitions, while Successor increasingly organizes it around long-term behavior. Their composition guides policy improvement through one physical prediction and one behavioral summary, without long imagined rollouts. In 20-second motion-tracking experiments, BeJEPA with online action refinement using Dynamic outperforms all evaluated baselines in global body-position error, reducing it on AMASS by 61.7% relative to BFM-Zero with world-pose inputs and 58.7% relative to SONIC. Physical deployment on a Unitree G1 demonstrates joint attainment of posture, world position, and heading goals, recovery from the ground, and closed-loop return after disturbances and goal changes. Project page: https://ylyem9x.github.io/BeJEPA-project.

0.27

Robust Design of Sensing-Assisted Secure Transmission for RIS-Aided ISAC System under Eavesdropper 3D Positioning Uncertainty

eess.SP ·2026-10-16 ·Jie Li, Zhendong Li, Wen Chen, Wei Xu, Ning Wang
benchmarktheory
Abstract · 12 pages, 7 figures

Localization errors in estimating the position of a potential eavesdropper (Eve) pose a critical challenge to secure transmission in reconfigurable intelligent surface (RIS)-assisted integrated sensing and communication (ISAC) systems. To address this issue, this paper proposes a sensing-assisted robust secure transmission framework that exploits ISAC sensing to characterize the positioning uncertainty induced by these errors. The RIS is partitioned into a reflection region and a sensing reception region, supporting sensing and communication through spatial separation. By jointly exploiting observations collected at the base station (BS) and the RIS sensing reception region, maximum likelihood estimation (MLE) is employed to estimate the arrival angles, from which Eve's three-dimensional (3D) position is obtained through cross localization. The resulting localization Cramer-Rao lower bound (CRLB) is then used to construct an ellipsoidal uncertainty region around the estimated position. A two-stage optimization strategy is developed. In Stage I, the sensing beamformer and RIS reflection coefficients are jointly optimized to improve localization accuracy. In Stage II, the resulting positioning uncertainty is incorporated into a robust secure transmission design, where the communication beamformer and RIS reflection coefficients are jointly optimized to maximize the worst-case secrecy rate. Simulation results demonstrate that the proposed approach outperforms benchmark schemes without RIS assistance and with random RIS configurations in terms of secrecy performance. The proposed framework effectively mitigates secrecy performance degradation caused by localization errors in Eve's position.

0.26

Reason-guided Human Reward Modeling for Preference-based Reinforcement Learning

cs.RO cs.AI ·2026-10-16 ·Anshul Nayak, Shahil Shaik, Jonathan Smereka, Yue Wang
Abstract

Reinforcement learning (RL) policies optimized for task rewards may produce behaviors that are efficient but misaligned with human expectations, safety preferences, or intent. This paper presents a reason-guided preference-based reinforcement learning (PbRL) framework that learns a human reward model from trajectory-level feedback. In addition to preference labels, the framework incorporates reason tags explaining why a behavior is preferred, enabling the learned reward to capture both human choices and their underlying rationale. Compared with preference-only reward learning, the reason-guided formulation provides additional supervision for learning human-aligned rewards and improves the interpretability of the resulting policy. The learned human reward is combined with the environment reward during policy optimization, allowing the agent to improve task performance while producing more interpretable, human- aligned, and preference-sensitive behavior

0.26

Lyapunov Exponents in Power Systems: A Survey of Theory and Applications in Stability, Control, Forecasting, and Protection

eess.SY ·2026-10-16 ·Mohammad Almomani, Venkataramana Ajjarapu
survey
Abstract

Lyapunov exponents (LEs) are fundamental metrics from nonlinear dynamical systems theory that quantify the average exponential rate of divergence or convergence of nearby trajectories in phase space. Over the past three decades, LEs have emerged as powerful tools across a wide range of power system problems, including transient and voltage stability assessment, chaos detection and mitigation, short-term load and renewable energy forecasting, and fault diagnosis. This paper presents a comprehensive review of these applications, covering more than 165 publications spanning from the early demonstrations of chaos in simplified machine models (1990s) to the most recent data-driven and machine-learning-augmented approaches applied to inverter-based resource (IBR)-dominated grids (2024--2026). The review is organized into four principal application domains: (i) stability analysis and real-time monitoring, (ii) chaos characterization and control, (iii) forecasting and predictability analysis, and (iv) protection, fault diagnosis, and condition monitoring. For each domain, we trace the evolution of methods, compare model-based and data-driven LE estimation strategies, identify key benchmarks and validation gaps, and outline the most pressing open research problems.

0.25

Beyond Success Rate: Interventional Evaluation of Memory in Vision-Language-Action Systems

cs.RO cs.LG ·2026-10-16 ·Keyuan Luo, Yi Xie, Wanyun Ling, Yuanzhe Li, Ziyue Li +1 more
Abstract · 41 pages, including references and appendices

When we say that a robot uses memory, does that memory actually play a role? Success-rate ablations alone cannot establish the use of task-relevant facts: they change information and computation together, while aggregate rates can hide offsetting gains and losses. We propose an Interventional Memory Evaluation Protocol (IMEP) for vision-language-action (VLA) systems to address four questions: (1) Does memory provide a net benefit? (2) Under what conditions does memory help? (3) Do changes to memory alter task outcomes? (4) Are these effects specific to task-relevant facts? At the retrieval interface, we compare target-fact substitutions with fact-preserving replacements, minimal numerical perturbations, and exact replicas, while assessing decision correctness separately from motor execution success. We evaluate within-episode, cross-episode, and agentic memory. In cross-episode learned-bank experiments, a one-unit-in-the-last-place (ULP) intervention, including its installation procedure, changes 11.9% of seed-paired outcomes, while changing key facts produces no clear additional effect over fact-preserving replacements. To examine when memory helps, we introduce ChainBench, where correct decisions require historical information unavailable in current observations. The full memory configuration raises task success from 39.1% to 74.3%, matching the oracle-decision condition under shared recorded executions. These results show that correct history can support robot decisions, but establishing whether a memory system realizes this benefit requires fact-specific evidence beyond aggregate success rates.

0.25

MOSS: Multimodal Origami Swamp Robot System

cs.RO ·2026-10-16 ·Tian Lan, Atar Babgei, Luca Romanello, Mirko Kovac, Sophie F. Armanini
Abstract · 8 pages, 8 figures, submitted to ICRA 2027

Robotic systems for environmental monitoring and ecological surveying in wetlands and swamps must operate across diverse terrestrial and aquatic conditions. Conventional fixed-geometry wheels cannot adapt their contact geometry to these varying conditions. This work presents MOSS (Multimodal Origami Swamp Robot System), a compact amphibious robotic platform with 3D-printed origami wheels. The compact crank mechanism switches the wheel between contracted and expanded configurations, changing its width, body diameter, and fin orientation simultaneously. The contracted configuration is better suited to efficient rolling on firmer ground, whereas the expanded configuration provides greater contact and terrain engagement on softer surfaces. MOSS was evaluated on soil, sand, mud with varying water content, and water, in addition to in-motion morphing and terrestrial-aquatic transition tests. The more energy-efficient wheel configuration depended on terrain. The contracted configuration achieved a lower cost of transport on firmer soil and mud, whereas the expanded configuration performed better on wetter mud, sand, and water, reducing the cost of transport by up to 29%. Expanded wheels also reduced slip in wet mud and reduced both slip and hub sinkage in sand. MOSS successfully completed configuration changes in $1.95 \pm 0.11$ s and demonstrated continuous locomotion across soil-mud-water transitions. Outdoor trials further demonstrated operation across natural deformable terrain and water. These results show that adaptive wheel geometry provides terrain-dependent locomotion benefits and enables a single compact platform to traverse substantially different terrestrial and aquatic conditions.

0.25

Sample-Efficient Reinforcement Learning for Policy Improvement via Flow-Reversed Action Anchors

cs.RO cs.AI ·2026-10-16 ·Carl Qi, Weijun Dong, Xiaojie Wang, Peiyi Chen, Rahul Sajnani +6 more
Abstract

Generalist robot policies offer strong behavioral priors, but achieving deployment-level reliability and throughput remains difficult. Reinforcement learning (RL) can improve these policies, yet its effectiveness is constrained by the cost of online interaction. Existing methods adapt frozen flow-matching policies by treating noise space as a continuous action space. We show that this behaviorally unstructured space makes exploration inefficient, often preventing these methods from extracting meaningful learning signals on challenging tasks. We introduce AnchorQ, which enables coarse-to-fine policy improvement through a structured, refinable latent action representation. At each decision step, AnchorQ uses flow reversal to map reference behaviors to state-conditioned latent anchors, samples local particles around those anchors, and applies value-guided sampling to explore promising refinements. Our analysis shows that anchors reduce identification complexity at the cost of support bias; particle refinement trades this bias against action-class complexity; and critic accuracy determines the reliability of KL-regularized, value-guided selection. Across simulated and real-world manipulation tasks, AnchorQ outperforms state-of-the-art post-training baselines in sample efficiency, delivering a 37% improvement in real-world success rate and 97% higher simulated task throughput under the same interaction budget. Additional visualizations are available on our website https://anchorq-rl.com/.

0.25

An Efficient Blind Detection of OFDM Signals with Unknown Parameters

eess.SP ·2026-10-16 ·Ata Bartu Soyuer, Gokhan M. Guvensen
Abstract · Submitted to IEEE Transactions on Aerospace and Electronic Systems (TAES) for possible publication

This study investigates efficient detector designs for the blind detection of orthogonal frequency-division multiplexing (OFDM) signals with unknown power, bandwidth, and carrier frequency in wideband spectrum sensing. We first describe the received signal model and conduct an asymptotic covariance analysis to obtain a tractable discrete-time representation. Based on this model, generalized likelihood ratio test (GLRT) detectors are derived for both known and unknown noise power cases in a single-antenna receiver setting. The proposed framework is subsequently extended to multi-antenna receivers, and the corresponding multi-antenna GLRT detectors are developed for both known and unknown noise power cases. Importantly, using the cumulative-sum (CUSUM) operations, the resulting detectors reduce to an efficient one-dimensional (1D) search in both single-antenna and multi-antenna settings. We also provide theoretical analysis for the conventional energy detector (Conv-ED) and the hypothetical uniformly most powerful (UMP) test to reveal the impact of system parameters on detection performance. In addition, computational complexity analysis is presented for both the proposed and conventional detectors. Finally, comprehensive simulation results are provided to evaluate the probability of detection. The results demonstrate that the proposed detectors achieve performance close to that of a genie-aided detector while significantly outperforming conventional detection approaches.

0.24

Learning When to Listen: Physics-Informed Synchronization in Molecular Channels

eess.SP ·2026-10-16 ·Osman Tugay Basaran, Saswati Pal, Falko Dressler, Jorge Torres Gómez
Abstract

Recent advances in chemical computing create new opportunities to realize biocompatible communication devices for application in precision medicine. Implanted microscopic transmitters and receivers can exchange information using molecules that diffuse through a biological medium, enabling them to form a network and jointly coordinate sensing and actuation operations. However, randomness in molecular propagation causes severe signal distortion, and receivers have limited computational capacity to compensate. The receiver design must therefore combine robustness to diffusion-induced distortion with computational architectures amenable to implementation in the chemical domain. To meet these requirements, we propose a physics-informed reinforcement learning (RL) framework implemented with the reservoir computing (RC) architecture. We study the task of synchronizing the receiver's clock with a sequence transmitted through a molecular communication (MC) channel. The physics part approximates the maximum-likelihood function synchronization objective, while the other learns a timing policy using this objective as its reward. Simulation results yield clock misalignments above half a second, compatible with molecular transmissions operating on the timescale of seconds.

0.24

SimpleVLN: Towards Simple and Frontier Vision-Language Navigation

cs.RO ·2026-10-16 ·Lingyi Hong, Weiren Cai, Jiayu Chen, Xinyu Zhou, Bailin He +5 more
Abstract · homepage: https://simplevln.github.io/

Vision-and-Language Navigation (VLN) requires embodied agents to ground language in visual observations and make long-horizon decisions. Existing methods often address these demands through task-specific architectures. As multimodal large language models (MLLMs) become increasingly capable, an important question remains: Have we truly unlocked the potential of native MLLMs for VLN before introducing task-specific architecture? We present SimpleVLN, an architecture-preserving framework developed through studies of navigation formulation, data construction, and optimization. At the formulation level, SimpleVLN combines single-turn decision inputs with fixed-budget global-local memory to efficiently use navigation history without unbounded context growth, while action chunks enable efficient output of multiple primitive actions per model invocation. For data construction, we combine expert demonstrations with policy-dependent DAgger experience and develop a scalable synthesis pipeline that generates large-scale tasks for reinforcement learning. At the optimization level, a progressive IL-DAgger-RL pipeline advances from action imitation to task-level outcome optimization. Comprehensive experiments validate the effectiveness of these three aspects. We instantiate SimpleVLN across multiple MLLM backbones and model sizes, demonstrating its generality. SimpleVLN achieves 76.7% and 80.1% success rates on the R2R-CE and RxR-CE splits, respectively, substantially outperforming existing methods. These results show that the navigation capabilities of native MLLMs can be effectively unlocked without task-specific architectural designs.

0.24

Temporal vs Spectral Resolution in CSI Reporting for Wi-Fi ISAC

eess.SP ·2026-10-16 ·Abolfazl Zakeri, Sumit Roy, Nhan Thanh Nguyen, Markku Juntti
Abstract · Under review for publication

CSI feedback is fundamental to proposed uses of Wi-Fi sensing such as human activity recognition (HAR); but it also constitutes significant overhead as resources must be allocated away from data communications. Wi-Fi CSI extraction can be explored along two dimensions in IEEE 802.11bf: reporting intensity, i.e., temporal resolution, and number of reported subcarriers, i.e., spectral resolution. We explore periodic and IEEE 802.11bf threshold-based reporting at matched rates, and uniform decimation and block-mean grouping at matched report sizes. We further propose reporting the mean CSI accumulated since the last report. Using the 802.11bf report size as the metric, we evaluate the accuracy-cost tradeoffs on three public datasets with our proposed model, a convolutional neural network combined with a gated recurrent unit and temporal pyramid pooling. The proposed periodic reporting with mean payload and block-mean grouping improves accuracy by up to 53 percentage points at equal CSI overhead cost. Temporal resolution is generally more cost-effective than spectral resolution. With only at most 7.7% of full-CSI feedback cost, the proposed scheme achieves 96-98% of full-CSI accuracy.

0.23

Stochastic Shockwave Dynamics in First-Order Traffic Flow

eess.SY physics.soc-phmath.APmath.PR ·2026-10-16 ·Yanlin Zhang, Alireza Talebpour
unusual
Abstract

Moving congestion fronts vary across realizations even when the traffic-flow law is fixed. This paper develops a generator-based formulation of the Lighthill--Whitham--Richards (LWR) model with random initial density and a deterministic fundamental diagram. Randomness enters through the initial spatial density field; every realization follows the same conservation law. For a smooth strictly concave fundamental diagram, we adapt established one-sided spatial Markov closure to traffic variables under the stated spatial-process and regularity assumptions, and derive a kinetic balance for physically normalized front intensities. Evolving the spatial generator provides density laws and congestion-front statistics without solving a road-grid problem for every realization. The formulation is continuous in physical space and time, while its numerical implementation approximates density state and integrates physical time. Four experiments distinguish deterministic consistency from stochastic comparison. Riemann and signal examples recover independently constructed exact LWR solutions. Smooth strictly concave approximations approach the deterministic triangular-diagram signal solution over the tested finite horizons; this does not establish stochastic closure at contacts or the capacity kink. A matched Monte Carlo Godunov ensemble evaluates front counts, sizes, mergers, and density distributions. The evolved generator also yields an instantaneous travel-time distribution for a frozen density field, compared with the same functional of the matched Godunov profiles. This quantity is not experienced travel time through evolving traffic. Resolution checks identify remaining normalization and small-front-loss effects. The contribution is direct evolution and interpretation of traffic-state and front distributions under a fixed first-order traffic law.

0.23

Integrated Sensing and Communication in 3GPP: Standardization, Performance Evaluation, and Experimental Validation Toward 6G

eess.SP ·2026-10-16 ·Gosan Noh, Gebremeskel Seiley Abraha, Joshua Yabana, Nathaniel Foli, Michael Ankomah Asare +8 more
Abstract

Integrated sensing and communication (ISAC) is evolving from a primarily physical-layer research into a standardized cellular-network capability spanning channel modeling, radio procedures, network architecture, and the delivery of sensing services. We review this progression within the Third Generation Partnership Project (3GPP), from Release 19 service requirements and target-aware channel modeling to the Release 20 New Radio (NR) sensing evaluation framework, radio access network procedures, and system architecture, and discuss future directions toward 6G. We then present two 3GPP-aligned simulation case studies to demonstrate the corresponding evaluation approaches. The first case study compares hard-decision, soft-decision, and hybrid fusion for unmanned aerial vehicle (UAV) sensing using multiple transmission and reception points (TRPs). Detection, false alarms, and conditional position and velocity accuracy are evaluated using receiver-derived angle estimates and separately calibrated method-specific thresholds. The second case study compares ideal and practical sensing-assisted predictive handover with conventional handover, while accounting for sensing error and information age. We further present a testbed-based measurement study that characterizes monostatic and bistatic UAV radar cross sections across frequencies, platforms, and observation angles and provides empirical support for correlated target-response modeling. Finally, we discuss open challenges and future research directions in multi-node fusion, resource and interference management, network evolution, security, privacy, and reproducible evaluation. Together, these components connect calibrated channel models, link- and system-level simulation, real-time prototypes, controlled measurements, and field trials through a staged evaluation roadmap.

0.23

Learning Robot Metareasoning: Adaptive Computation Budgets for Real-Time Control

cs.RO ·2026-10-16 ·Lorenzo Magnino, Antonio Marino, Michael Amir, Daigo Shishika, Amanda Prorok
Abstract

Planning while moving faces an historical trade-off between deliberation and reactivity. A planner running concurrently with execution needs time to compute, so the robot must commit to an open-loop action chunk for that duration: a longer commitment buys a better plan at the cost of feedback frequency. Existing controllers fix this action commitment, and consequently the computation budget, before deployment. However, the optimal computation time is highly state-dependent (due to factors such as unmodeled dynamics, safety constraints, and marginal optimisation gains), i.e. dynamic, therefore offline static tuning risks leaving systems blind during critical maneuvers or unable to find global effective solutions. We cast the identification of the optimal computation budget as a semi-Markov decision process (SMDP) in which a single decision defines both the robot action commitments and the time allocated to the solver. To solve this, we train a reinforcement-learning policy conditioned on the current state, the executing plan, and inexpensive planner statistics. Across ground-robot navigation, agile quadrotor flight, and contact-rich manipulation, the learned policy changes its budget with the demands of the scenario, discovering efficient reasoning patterns that outperform fixed-budget baselines. https://lorenzomagnino.github.io/learning-robot-metareasoning-review/

0.23

Rollbot 2: Agile Locomotion by a Single-Actuator Spherical Robot

cs.RO ·2026-10-16 ·Jingxian Wang*, Chen Yu*, Sam Kriegman, Michael Rubenstein
benchmark
Abstract

Reducing actuator count simplifies robot hardware but also removes independent inputs for maneuvering. The original Rollbot steered by modulating a quasi-stable helical orbit, but moved slowly and relied on external motion capture. Rollbot 2 preserves the single-actuator design while its redesigned mass distribution enables near-straight rolling and more direct travel toward commanded paths and beacon targets. Onboard inertial and ultra-wideband sensing support learned steering, beacon seeking, and waypoint following through one motor-velocity command. Rollbot 2 autonomously approaches UWB beacons without motion capture and completes a 468~m manually commanded indoor route at 0.87~m/s. On the shared 4.8~m ``N'' waypoint benchmark, all 16 trials complete the path with a mean traversal time of 10.8~s and mean time-averaged deviation of 7~cm. Its fastest traversal takes 8.0~s, over 70 times faster than the original Rollbot's 564~s run despite minor protocol differences. These results show how mechanical redesign, onboard sensing, and learning-based control improve the agility and autonomy of a single-actuator spherical robot.

0.22

A Koopman--Polynomial Chaos Framework for SMPC of Nonlinear Stochastic Processes

eess.SY ·2026-10-16 ·Wallace Gian Yion Tan, Efstathios Iliakis, Liang Wu, Jan Drgona, Richard D. Braatz
published
Abstract · A preliminary version of this work will be presented at the 2026 IEEE Conference on Decision and Control (CDC)

This paper develops a polynomial parametric Koopman operator (PPKO) framework for stochastic model predictive control (SMPC) of nonlinear systems with parametric uncertainty and time-varying stochastic disturbances. The Koopman operator is parameterized by polynomial chaos expansions (PCEs), enabling data-driven nonlinear dynamics to be integrated with PCE-based uncertainty propagation and control. We show that the polynomial parameterization introduces no irreducible approximation gap beyond PCE truncation. We then derive a condensed deterministic SMPC reformulation that handles both time-invariant parametric uncertainty and time-varying disturbances while retaining an online optimization dimension that depends only on the control horizon and input dimension, independent of the lifted-state and PCE dimensions. The resulting formulation yields convex programs with expectation and second-moment constraints that can be solved using standard optimization methods. Numerical studies on the Duffing oscillator and the electrothermal cooling system demonstrate the accuracy, scalability, and closed-loop performance of the proposed framework.

0.22

Joint Subcarrier and Power Allocation for Localization-Oriented Multi-Carrier Underwater ISAC Systems

eess.SP ·2026-10-16 ·Wei Huang, Sitong Liu, Christos Masouros, Hao Zhang, T. Aaron Gulliver +1 more
Abstract

Multi-carrier underwater integrated sensing and communication (ISAC) realizes concurrent acoustic communication and localization, yet the frequency selectivity of the acoustic channel creates uneven positioning capabilities across subcarriers. This paper proposes a localization-oriented subcarrier and power allocation scheme for multi-anchor underwater ISAC systems. We build a unified orthogonal subcarrier multiplexing framework and derive closed-form Cramer-Rao bounds (CRBs) under frequency-selective channels with colored noise. We derive a subcarrier-level localization information contribution from the time of arrival Fisher information, which enables localization-oriented resource allocation. We derive several theoretical properties regarding the distance and frequency dependence of the localization information metric, and establish a distance-frequency matching principle. These results provide a theoretical foundation for localization-oriented resource allocation and guide the design of a low-complexity greedy algorithm that improves positioning accuracy. Specifically, the algorithm iteratively allocates subcarriers to minimize the residual positioning-error objective, thereby improving positioning accuracy while satisfying communication quality-of-service (QoS) constraints. Simulations verify that our method yields much lower positioning error bound than uniform, random and high-frequency-first subcarrier allocation in symmetric or asymmetric topologies while guaranteeing communication QoS, achieving superior communication-localization tradeoffs.

0.22

PASS: A Trustworthy Precision Approach Simulation System for CAT III Integrity Algorithm Development

cs.RO eess.SP ·2026-10-16 ·Lang Zhu, Xin Zhang, Yulong Sun, Nan Liu, Jiahe Shen +2 more
Abstract

For fixed-wing aircraft, reliable localization during approach and landing is critical to flight safety. Visual--inertial navigation is promising, but aircraft navigation also requires integrity: trustworthy outputs and timely warnings when they should not be used. Real approach data are costly to collect, while many visual errors and failures required for integrity algorithm development are difficult to reproduce safely. We present PASS, a trustworthy Precision Approach Simulation System for CAT III integrity algorithm development. In a unified temporal and geometric framework, PASS integrates configurable airport environments, prescribed trajectories, inertial simulation, calibrated stereo images, and continuous runway-relative ground truth. It supports navigation and integrity research through controllable sensor errors and environmental variations, computable visual projection errors, deterministic runway geometry priors, and condition-labeled data for statistical error modeling. We further propose Sim-to-Real Scene Trustworthiness (SRST), a measure based on DINOv2 feature distributions and UMAP, to quantify sim-to-real closeness. Experiments with representative stereo--inertial estimators demonstrate the usability of the generated sequences and quantify localization errors across algorithms and environments. PASS provides a controllable foundation for developing and evaluating visual--inertial navigation and integrity algorithms for fixed-wing precision approaches up to CAT III requirements.

0.22

ACTS: Learning Visuo-Tactile World Models from Human Physical Interaction

cs.RO ·2026-10-16 ·Yuxiang Ma, Weihan Xu, Pengfei Ye, Anna Mei, Yuxin Ray Song +3 more
Abstract · Project website: https://acts-anon.github.io/

Abstract Action-conditioned world models for contact-rich manipulation require data that captures how contact forms, slides, and releases under varying force. Such data are slow to collect through robot teleoperation, which often lacks force feedback and is constrained by gripper geometry. We ask whether these dynamics can instead be learned from human physical interaction and transferred to a robot. We present ACTS, a visuo-tactile world model trained only on human interaction data. Operators manipulate handheld GelSight sensors while we record third-person and wrist video, tactile images, 6-DoF sensor motion, and tactile-derived normal force, yielding React, a 4.9-hour bimanual dataset across three objects. We represent actions at the tactile sensor as motion augmented by force change, providing a shared action space for human collection and robot execution. ACTS jointly predicts future scene, wrist, and tactile observations. On held-out human data, ACTS improves scene-view prediction by 0.7-2.0 dB PSNR over the stronger of two visuo-tactile world models trained from scratch on React. Without fine-tuning, in the robot teleoperation rollouts we inspected, predicted tactile imprints still appear and disappear with recorded contact. Visual transfer is less robust: camera predictions exceed frame copying on PushT, are mixed on Rope, and fall below it on Motherboard, while wrist views can blur or hallucinate structure. These results suggest that contact dynamics transfer more reliably than visual appearance from human to robot.

0.22

ORD-WM: Learning Object-Centric Rigid Dynamics for 3D World Modeling in Robotic Manipulation

cs.RO ·2026-10-16 ·Yueyang Weng, Xiaopeng Zhang, Yuxi Ou, Yongjin Mu, Chengzhi Zhao +1 more
Abstract

World models provide robots with a predictive mechanism for reasoning about how actions change object states in manipulation tasks. However, existing image-based and dense 3D world models do not explicitly impose object-level rigid motion, leaving rigid-object geometry preservation to be learned implicitly and making multi-step rollouts susceptible to shape drift or object-level motion inconsistency. To address this, we propose ORD-WM, a 3D world model that learns object-centric rigid dynamics for robotic manipulation. ORD-WM learns a vector-neuron variational autoencoder supervised by signed distance fields, which encodes each object point cloud as a geometry-aware latent representation composed of 3D vector features, and uses an action-conditioned dynamics model to predict object-wise SE(3) motion increments. Rather than generating unconstrained next latent states, ORD-WM analytically applies predicted object-wise rigid transformations to the current latent states, separating persistent object geometry from pose changes. Experiments show that ORD-WM enables stable, geometry-preserving multi-step prediction and supports efficient real-time replanning for closed-loop model predictive control in simulated and real-world manipulation scenarios.

0.22

Directed-Hypergraph Analytics and Fragility Testing for Anti-Money Laundering

eess.SP ·2026-10-16 ·Run Li, Paul Laux
Abstract

Illicit activity in financial networks is difficult to detect. We exploit the idea that the chance of a transaction being illicit depends partly on the concurrent positions and recent actions of counterparties, as well as the higher-order organization of fund flows. We combine hypergraph features and machine learning to demonstrate practical ef fectiveness. Specifically, we use the Elliptic++ Bitcoin dataset to show that graph and directed-hypergraph-based features improve illicit-address classification. The suc cess of the predictive model then allows us to examine potential money-laundering chains are structurally fragile. Several illicit-to-licit transaction motifs with unknown intermediary addresses specified, and the fragility of these motifs is evaluated through progressive removal of the highest-risk intermediaries. Surprisingly, these motifs are fragile, suggesting that money laundering could be disrupted via the identification and exclusion of owners of a limited number of crypto wallets.

0.22

Packet Loss Analysis Under Doppler Effects and Reinforcement-Learning-Based Resource Allocation for Direct-to-Satellite LoRaWAN

eess.SP ·2026-10-16 ·Alireza Maleki, Ebrahim Bedeer, Robert Barton
Abstract

Long range wide area network (LoRaWAN) has a long communication range and low power consumption, making it a potential technology for direct-to-satellite (DtS) Internet of Things (IoT) connectivity. However, packet collisions and Doppler impairments brought on by satellite motion can greatly increase the packet loss ratio (PLR) in large-scale DtS-IoT networks. In this paper, we study resource allocation in DtSIoT LoRaWAN by first developing an analytical PLR model that simultaneously takes Doppler impairments and packet collisions in the temporal and frequency domains into account. In order to maximize resource allocation, the generated PLR is then used as the reward function for two reinforcement learning (RL) schemes: deep Q-network (DQN) and proximal policy optimization (PPO). Simulation results validate the analytical PLR model against Monte Carlo simulations and demonstrate significant improvements over random resource allocation. At a target PLR of 3*10^-2, while random allocation supports only 100 end devices (EDs), PPO and DQN support approximately 650 and 690 EDs, respectively, representing nearly a seven-fold increase. Furthermore, for 800 EDs, PPO and DQN increase the total number of transmitted packets by 106.7% and 88.0%, respectively, compared with random allocation.

0.22

Beyond Latent Personalization: Integrating HRV-Conditioned Personalization via LLM for Interpretable Stress Detection

eess.SP cs.LG ·2026-10-16 ·Wei-Hsiang Ma, Ya-Tse Wu, Hsing-Hang Chou, Chi-Chun Lee
Abstract · Submitted to ICASSP 2027

Stress detection based on heart rate variability (HRV) has received increasing attention. However, existing personalized approaches integrate personal attributes with HRV through latent representations, limiting the interpretability of their interactions. In this work, we propose an LLM-enhanced personalized stress detection framework that integrates HRV features and personal attributes through LLM-based semantic interpretation. The proposed framework achieves a 29.9% relative MCC improvement over the strongest baseline on TILES-2018 while providing human-readable descriptions of individual physiological responses. The results highlight the potential of LLM-based semantic interpretation for achieving both effective and interpretable personalized stress detection.

0.21

Entropy-Based Wireless Technology Recognition

eess.SP ·2026-10-16 ·Mohammad Alhellani, Eli De Poorter, Mamoun Guenach, Adnan Shahid
Abstract

Wireless Technology Recognition (WTR) enables identifying wireless technologies from received radio signals and is important for spectrum monitoring and management systems. Existing approaches either rely on handcrafted signal features with limited robustness or computationally intensive Deep Learning (DL) models operating on high-dimensional In-phase and Quadrature (IQ) samples or spectrograms. In this paper, we investigate a different hypothesis: wireless technologies can be distinguished through the temporal evolution of their local time-frequency complexity. To capture this behavior, we propose a lightweight WTR framework based on two-dimensional dispersion entropy. Entropy values are extracted from successive spectrogram regions to characterize the evolution of local spectrogram complexity throughout a transmission and summarized into a compact feature representation for classification using a Multilayer Perceptron (MLP). The framework is evaluated on seven wireless technologies, namely IEEE 802.11ah, IEEE 802.15.4 SUN-FSK, IEEE 802.15.4 SUN-OFDM, LoRa, IEEE 802.11n, LTE, and 5G NR, at Signal-to-Noise Ratios (SNRs) ranging from -20 dB to 40 dB. Experimental results show that the proposed approach consistently outperforms IQ-based and Short-Time Fourier Transform (STFT)-based Convolutional Neural Network (CNN) baselines, particularly at low and moderate SNRs. Averaged across all evaluated SNR levels, it achieves 82.73% accuracy, corresponding to improvements of 23.8% and 14.6% over the IQ-based and STFT-based CNN approaches, respectively. Furthermore, the approach maintains its recognition advantage with data-augmented training while requiring substantially shorter training times. The results demonstrate that temporal evolution of local spectrogram complexity provides a robust technology-specific fingerprint and an effective, computationally efficient representation for WTR.

0.21

MIXSim: A multiclass first order traffic model with composition-dependent dynamics for mixed-autonomy urban networks

eess.SY math-ph ·2026-10-16 ·Muhammad Haris, Claudio Roncoli
Abstract

This paper presents a novel multiclass first order macroscopic traffic model for mixed-autonomy in signalized urban networks, called MIXSim. Human driven vehicles (HDVs) and Connected and Automated Vehicles (CAVs) are represented as separate classes while origin-destination information is retained through commodity specific states. The local vehicle class composition and class specific driving behaviour characteristics determine the parameters of a composition-dependent triangular fundamental diagram. At signalized intersections, the composition of movement demand additionally determines the saturation flow rate, while the realized movement flow is constrained by demand, effective green time, and downstream receiving capacity. Numerical experiments investigate different CAV penetration levels and CAV time gaps, together with alternative signal control strategies. The numerical experiments illustrate how changes in CAV penetration and CAV time gap affect network performance under the tested conditions, and demonstrate the use of MIXSim for comparing signal control strategies across different mixed traffic regimes.

0.21

F-LIVO: A Feature-Level Geometric Measurement Framework for LiDAR-Inertial-Visual Odometry

cs.RO ·2026-10-16 ·Xinye Dai, Yapeng Wang, Shujie Zhou, Zihao Wang, Shiqi Zheng +4 more
code/data
Abstract · 19 pages, 20 figures

Measurement-level LiDAR-Inertial-Visual Odometry (LIVO) can achieve high accuracy, and its practical resource requirements depend on the adopted association and map-maintenance strategy. Persistently attaching dense photometric patches or appearance attributes to mapped LiDAR points can introduce additional storage, association, and update costs. This paper presents F-LIVO, a filter-based LIVO framework that couples a visual-inertial subsystem and a LiDAR-inertial subsystem through feature-level point-to-plane constraints between visual points and LiDAR planes. To preserve cross-modal coupling when visual triangulation is weak, we further recover visual feature depth from the LiDAR map through local plane fitting and ray-plane intersection. In the LiDAR update, the covariance of each point-to-plane residual is adaptively scaled according to local planar reliability, thereby reducing overconfident updates on poorly conditioned surfaces. Because the proposed method relies on sparse geometric features rather than persistent dense photometric map attributes, it retains a compact map representation. Experiments on multiple public datasets show competitive, sequence-dependent accuracy. Across the evaluated degradation cases, F-LIVO maintains continuous estimation with comparatively low accumulated end-to-end drift. Under the evaluated resource configuration, it achieves the lowest mean and peak central processing unit (CPU) usage, the lowest peak resident set size (RSS), and the lowest RSS growth rate among the evaluated implementations, while R2LIVE retains a lower mean RSS. The source code will be made publicly available upon acceptance to facilitate reproducible research.

0.21

Anytime-Valid Certification of Robustness Margins for Safe Control via E-Values

eess.SY ·2026-10-16 ·Panagiotis S. Trakas, Omid Mirzaeedodangeh, Lars Lindemann
Abstract

Robust control uses the error between a known model and the true system as a robustness margin. However, this error is usually unknown: conservative estimates restrict feasibility and performance, while underestimates can result in violations of the control property. We estimate this robustness margin from online data using e-values, a statistical tool for hypothesis testing, in the setting of discrete-time control barrier functions (CBFs). Here, we measure the robustness margin as the difference between the observed CBF and the model-based CBF prediction. We then design a robust CBF controller in three steps by: (1) selecting a set of candidate margins, (2) obtaining validity certificates for these candidates via sequential tests based on e-values, and (3) updating the CBF controller with the smallest certified candidate. The deployed tests accumulate evidence against the hypothesis that an exceedance risk, the probability of the robustness margin exceeding the candidate margin, remains at or above a pre-specified failure probability $α$. Our tests are anytime valid and thus enable the use of certified candidates in the CBF controller, guaranteeing that the probability that the exceedance risk of a certified margin exceeds $α$ is at most a user-chosen confidence $δ$. We validate our approach on navigation problems across different environments with visualizations in NVIDIA Isaac Sim.

0.20

GRASP: Global-Local Representation Alignment and Spatial-Physical Grounding for Bimanual Robotic Manipulation

cs.RO ·2026-10-16 ·Zeyuan Ren, Guanxun Zhao, Qi Li, Hao Wang, Wenlong Liao +13 more
Abstract · 8 pages, 4 figures

Vision-Language-Action (VLA) models that directly regress joint-space actions avoid the deployment-time cost of inverse kinematics. However, their training objectives remain misaligned with task success, as joint-space losses ignore the unequal physical consequences of joint errors and visual conditioning lacks explicit focus on hand-object interaction regions. To address these issues, we propose a training-phase prior injection framework for fine-grained bimanual manipulation. First, a physics-grounded flow-matching loss maps the implicit clean-action estimate at every noise level through differentiable forward kinematics (FK) and supervises the resulting Tool Center Point (TCP) pose and inter-arm relative pose, with a bounded noise-level weighting that compensates gradient decay. Second, we introduce a global prediction query group distilled from a frozen DINOv3 teacher through future-frame feature prediction, together with a local interaction query supervised by Grounding DINO at the attention level. Both teachers are removed at inference, and the deployment pipeline is that of a standard joint-space flow policy, adding only the learned query pathway. Simulation and real-robot experiments on long-horizon bimanual tasks show that the two priors are complementary across different deployment settings. The full model excels in in-distribution deployment, while the distilled query branch is preferable under strong visual randomization.

0.20

Characterization, ageing and post-mortem dataset of commercial 58Ah prismatic NMC cells

eess.SY ·2026-10-16 ·Róbinsón Medina, Alessio A. Lodge, Feye S. J. Hoekstra, Steven Wilkins, Damian Cupid +8 more
Abstract

This paper provides an extensive dataset for commercial 58 Ah prismatic NMC cells. The dataset comprises three types of data: characterization, ageing, and cell teardown. Characterization data include tests such as GITT, HPPC, and EIS, among others. Ageing data were obtained from 41 cells subjected to 38 distinct ageing conditions, including application-surrogate drive cycles. Recurring reference performance tests and EIS measurements track the evolution of each cell. Teardown data were obtained from pristine and aged (post-mortem) cells. They include material analyses and half-cell measurements. This single dataset supports analytics, model development (both machine-learning and physics-based), benchmarking, and techno-economic analysis. Several example analyses are carried out to show the usability of the data. For instance, the paper shows how lithium stripping can be detected from the voltage response of a short discharge applied after each charge. The data and the processing and analysis scripts are openly provided alongside this paper.

0.20

Pilot-Based Sensing With OFDM and AFDM: A Resource-Matched Performance Comparison

eess.SP ·2026-10-16 ·Siyi Tang, Yin Xu, Aimin Tang, Dazhi He, Wenjun Zhang
Abstract

As a waveform candidate for next-generation wireless systems, affine frequency division multiplexing (AFDM) has often been compared with orthogonal frequency division multiplexing (OFDM) under different configurations, while a systematic comparison of their pilot-based sensing performance remains lacking. This paper establishes a comparison with the same bandwidth, frame duration, pilot energy, and reserved-resource overhead. OFDM-random, OFDM-fixed-comb, and OFDM-shifted-comb pilots are compared with AFDM pilots surrounded by guard resources using aperiodic ambiguity functions without cyclic prefix/chirp-periodic prefix (CP/CPP) and cyclic ambiguity functions with CP/CPP, followed by target detection experiments. The results show that OFDM-fixed-comb exhibits low range sidelobes but periodic range ambiguity peaks, while AFDM and OFDM-shifted-comb effectively suppress zero-Doppler range ambiguities. Within the common unambiguous region, all pilot configurations for both waveforms exhibit similar detection thresholds in strong-target detection, while structured OFDM and AFDM pilot placements outperform OFDM-random in weak-target detection. At high mobility, the AFDM receiver using the discrete affine Fourier transform (DAFT) can resolve integer Doppler beyond the conventional slow-time maximum unambiguous velocity, outperforming the tested conventional OFDM and AFDM receivers at the increased search complexity. These results separate the effects of waveform, pilot placement, and receiver processing, providing a basis for AFDM and OFDM pilot-based sensing design.

0.20

SparseMax-dSTL: Sound Robustness with Horizon-Independent Gradient Allocation for Temporal Logic Trajectory Optimization

eess.SY cs.LOcs.RO ·2026-10-16 ·Sai Sandeep Damera, Ryan Matheu, John S. Baras, Calin Belta
theory
Abstract · 8 pages, 5 figures. This work has been submitted to the IEEE for possible publication

A gradient-based solver that enforces a Signal Temporal Logic (STL) specification uses a smooth approximation of its quantitative semantics, the robustness. In safety-critical trajectory optimization, this approximation must have two properties. First, it must be sound, that is, nonnegative only on trajectories that satisfy the specification. Second, at each temporal operator, the weight of its gradient on the violating samples must stay bounded away from zero, however long the horizon. The approximations in use, including the log-sum-exp, its sound variant, and the generalized-mean robustness, do not have both. We introduce lower approximations of the minimum and the maximum whose gradients are sparsemax projections, and we prove that the robustness built from them has both. It lies below the exact robustness and is therefore sound, and its error has a bound that is independent of the number of samples and known before optimization. The gradient of each approximation is exactly zero on every sample beyond a data-dependent threshold, so the weight on violating samples a gap beyond the others has a positive lower bound at every horizon. On a planar unicycle and on a seven-joint manipulator beside a person in simulation, the log-sum-exp approximation is positive on a violating trajectory. As the horizon of the manipulator's task grows from 10 to 80 s, the gradient weight of the sound variant on the violating samples falls from 0.22 to 0.04, while that of sparsemax stays at 1. In a constrained trajectory optimization of the unicycle, the constraint on the log-sum-exp robustness returns a violating trajectory, whereas the constraint on the sparsemax robustness returns a trajectory that satisfies the specification with the required margin.

0.20

GR-VIP: Geometric representation conditioning for precise visuomotor policy

cs.RO ·2026-10-16 ·Chanhwi Lee, Jiwon Seong, Eunje Choi, Hohyung Lee, Jongsoo Lee +4 more
Abstract · Submitted to IEEE Robotics and Automation Letters (RA-L). 8 pages, 5 figures. This work has been submitted to the IEEE for possible publicat…

Achieving precise terminal position and heading alignment remains a significant challenge for learning-based mobile robot control. Recent vision-based learning methods have shown promise for precise target-relative control and generalization to unseen environments with pretrained visual encoders, but general-purpose visual representations are not specifically designed to capture the fine-grained spatial differences between current and goal views required for precise control, and real-world demonstrations combining millimeter-scale terminal accuracy with zero-shot transfer across environments remain limited. To address these challenges, we propose Geometric Representation Conditioning for Precise Visuomotor Policy (GR-VIP), a goal-conditioned visuomotor policy that leverages a visual encoder pretrained for relative camera pose estimation. Rather than using the final relative-pose estimate, GR-VIP conditions action prediction on the dense relational features produced before pose regression, preserving fine-grained cross-view information for learning corrective actions from demonstrations. We evaluate GR-VIP on a real-world docking task and demonstrate high docking success rates and millimeter-scale terminal accuracy in both the training and unseen environments, without additional training for transfer. Furthermore, supplementary experiments illustrate the policy's ability to align with non-docking targets and complete docking under a tested low-light condition.

0.20

Simultaneous Force and Motion Control in Generalised Coordinates

cs.RO ·2026-10-16 ·Jon Woolfrey
Abstract · 8 pages, 5 figures, submitted to IEEE International Conference on Robotics and Automation (ICRA) 2027

Robot manipulators often come into contact with rigid objects when performing work. This imposes physical constraints on their motion. Simultaneous force & motion control was developed to regulate contact forces commensurate with these constraints, whilst projecting motion control in to the orthogonal, unconstrained directions. This is typically formulated in the task coordinates, i.e. operational space, then mapped to joint torques to command the robot. However, most constraints on the robot control, such as joint limits, collision avoidance, singularity avoidance, etc. are naturally expressed in generalised coordinates (joint space). Thus, formulating the control problem in task space makes it difficult to guarantee control feasibility. In this paper Gauss' principle of least constraint is used to solve the control as a quadratic programming (QP) problem in generalised coordinates. It is shown that the theoretical solution to this problem produces the correct constrained dynamics for accurate contact force control. Moreover, inequality constraints may be appended to ensure feasibility and safety. Simulation results were conducted in MuJoCo and compared to the operational space formulation. The proposed method is able to simultaneously maintain a desired contact force, minimise trajectory tracking error, and adhere to practical inequality constraints.

0.19

Multimodal Interaction Estimation for Contact-Aware Adaptive Impedance Control

cs.RO ·2026-10-16 ·Mateusz Jaszczuk, Paris Perdikaris, Nadia Figueroa
Abstract

With advancements in learning-based policies, robots are increasingly being deployed in human-centric environments, such as manufacturing floors and homes, where safe and precise operation requires compensating for unknown loads while remaining compliant during human interactions. Reliably distinguishing between these interaction types is challenging with a single sensing modality, as force signals cannot separate payloads from human pushes, and vision is susceptible to occlusions. We propose a task-agnostic interaction estimation and adaptive control framework that addresses this challenge by fusing force history and 3D point cloud through a transformer-based classifier. Trained with modality dropout for occlusion resilience and fine-tuned from simulation to hardware, the classifier achieves 91.89% accuracy and maintains above 90% under complete occlusion of one modality. The classifier drives a system identification framework that selectively compensates for unknown payloads and a variable-impedance controller that switches between adaptive and reactive modes, while simultaneously estimating interaction forces for downstream tasks, such as human intent estimation. We validate the framework under external perturbations and unknown loads across single-arm manipulation, deployment with an imitation-learning-based policy, intent-driven target switching, and zero-shot bimanual deployment on two Franka arms lifting a box of unknown mass, demonstrating generalization to closed kinematic chains and unseen setups, enabling safe and precise deployment.

0.19

Look Before You Act: Bridging Active Perception and Vision-Language-Action Models via Observation-Aligned Data Adaptation

cs.RO ·2026-10-16 ·Jiawei Geng, Zhenyang Liu, Yuzheng Wu, Yongchong Gu, Yanwei Fu
Abstract

Robots manipulating objects in clutter must decide not only how to act, but also where to look. Although active camera motion can expose an occluded target, it may move the robot into an initial observation-state distribution that differs from the manipulation policy's training distribution. We identify this problem as first-frame distribution sensitivity: under perturbed starts in LIBERO, naively prepending camera repositioning to pi_0.5 reduces success from 46.3% to 11.5%. We introduce ActiveAlign, a data-adaptation framework that trains a Vision-Language-Action (VLA) policy from the states at which active perception terminates. A learned viewpoint module first generates a camera trajectory. Offline, ActiveAlign uses closed-loop alignment to connect the trajectory endpoint to an existing demonstration, replays the remaining manipulation trajectory, and retains successful continuations for fine-tuning. At deployment, the same viewpoint module precedes the adapted VLA without modifying its architecture. Across 40 LIBERO tasks, ActiveAlign achieves 70.7% success with pi_0 and 72.7% with pi_0.5, outperforming their standard-training baselines by 40.0% and 26.4% respectively. Matched-state and data-matched controls support gains beyond prefix shortening and generic viewpoint augmentation; OpenVLA and GR00T N1 also show the same trend. Success rate under physical occlusion in real world increases from 35/90 to 83/90. These results show that active perception benefits manipulation when policy training is aligned with the observation and robot-state distribution induced by viewpoint selection.

0.19

Distributed Performance Boosting through Neural Control for Plug-and-Play Networks

eess.SY cs.LGmath.OC ·2026-10-16 ·Danilo Saccani, Leonardo Massai, Giancarlo Ferrari-Trecate
Abstract

We address the problem of improving the performance of pre-stabilized networked systems subject to plug-and-play (PnP) operations, namely additions or removals of subsystems. While distributed baseline controllers can preserve stability after a PnP transition, augmenting them with expressive neural policies can invalidate the formal stability guarantees. To overcome this issue, we build on recently proposed Performance Boosting controllers based on the Internal Model Control (IMC) principle, which reduce closed-loop stability to a finite $\ell_2$-gain requirement on an auxiliary operator. In the distributed setting, this operator is formed by interconnecting local neural operators that exchange information only between neighboring subsystems. We then derive free parametrizations of this interconnection, namely explicit maps from unconstrained trainable variables to interconnected neural operators satisfying a prescribed network gain bound for every parameter value. The obtained distributed neural controller can therefore be trained by unconstrained gradient-based optimization without stability constraints or projection steps, while retaining the prescribed gain bound at every training iterate. We develop two complementary parametrizations. A neighbor-based compositional parametrization uses local coupling bounds, so that after a PnP event only subsystems whose bounds change need to update their gain parameters. A network-norm parametrization instead uses a bound on the overall coupling matrix and can reduce conservatism, at the price of requiring network-wide information. A PnP DC-microgrid case study shows that the proposed neural layer improves voltage transients under topology changes and load variations while preserving nominal closed-loop stability.

0.19

Diversifying Moving Target Defense Against Adaptive Attacks Through Factorized Diffusion

eess.SY ·2026-10-16 ·Chenhan Xiao, Yang Weng
Abstract · 10 pages, 5 figures

Proactive defense offers a means to enhance power-system cybersecurity through controlled modifications of system configurations. One example is moving target defense (MTD), which uses distributed flexible AC transmission system (D-FACTS) devices to perturb transmission-line reactances against stealthy false data injection attacks. Existing MTD methods primarily rely on deterministic optimization, producing a single perturbation plan for each operating condition. However, repeated or highly correlated plans may expose predictable system configurations that attackers can learn and exploit. To address this limitation, this paper reformulates MTD as a conditional generative problem that produces diverse, secure, and physically feasible perturbation candidates. Learning such a distribution is challenging because MTD decisions jointly involve selecting branches, determining perturbation magnitudes, and coordinating adjustments across the network. To make this problem tractable, we factorize the decision distribution into three sequential conditional stages: branch selection, magnitude allocation, and continuous refinement. Graph-conditioned diffusion models are employed to capture the discrete and continuous decisions while accounting for network dependencies. Furthermore, a - objective is incorporated into diffusion training to favor candidates with desirable defense effectiveness and operational costs. Numerical studies on the IEEE 118- and 300-bus systems show that the proposed framework achieves AC feasibility comparable to that of optimization-based MTD baselines and higher - values, while reducing online computation time by an order of magnitude. The generated candidate pool remains effective under changing operating conditions and device availability, providing diverse high-value alternatives for flexible online cyber defense.

0.19

Not All Actions Contribute Equally: Learning Precise Progress Estimation from Imprecise Demonstrations

cs.RO ·2026-10-16 ·Huanqian Wang, Boshi Tang, Botong Xiao, Jie Zhang, Yang Yue +3 more
position
Abstract · 9 pages, 13 figures

Progress estimators play an essential role in providing dense reward signals for reinforcement learning in robotic manipulation, but accurately assessing task progress can be much challenging. Existing methods heavily rely on expert data and naively assume that all actions within a demonstration contribute to task success. This assumption often breaks down in real-world scenarios, especially in high-precision manipulation, where subtle geometric errors can lead to task failure. Therefore, these methods can erroneously interpret human mistakes as meaningful progress, further corrupting progress learning. We argue that a key limitation is the lack of a global prior regarding task evaluation criteria, which is crucial for distinguishing meaningful progress from human mistakes. To this end, we propose PRIME, a novel progress estimator capable of learning precise progress estimation from imprecise demonstrations by aggregating experience across trajectories. PRIME constructs a weighted graph over observed states, connecting states across trajectories based on spatial proximity and visual consistency. Edge weights encode spatial distances as transition costs, and shortest-path search approximates the geodesic distance to success. Changes in this distance naturally distinguish advancement, stagnation, and setbacks without assuming monotonic progress over time. Extensive experiments in both simulation and real world on precise manipulation tasks show that PRIME is consistently effective across diverse tasks, and robust to data with varying degrees of suboptimality. Notably, the advantage of PRIME gets more pronounced in higher-precision scenarios, yielding up to a \(5\times\) improvement in policy success rate over the SFT baselines on charger insertion.

0.19

Shape-Aware Neural Time Fields for Rigid-Body Motion and Manipulation Planning

cs.RO cs.LG ·2026-10-16 ·Yanming Zhou, Yuchen Liu, Ahmed H. Qureshi
Abstract

Rigid-body motion planning is fundamental to robotic transport and manipulation, where both object position and orientation determine whether a body can move through cluttered environments. This coupling makes planning chal- lenging because translation and rotation can be constrained very differently by surrounding geometry. We propose Shape- Aware Neural Time Fields, a physics-informed neural motion planning framework that learns a reusable travel-time, or cost- to-go, representation for rigid-body motion while explicitly accounting for these distinct geometric restrictions. The learned field captures local motion difficulty in translation and rotation and is paired with a sampling-based planner that balances progress toward the goal with the cost of reaching nearby poses. We evaluate the framework across 2D and 3D rigid- body planning tasks and non-prehensile manipulation through object pushing in diverse environments. Results show substantial improvements in planning success and inference efficiency over physics-informed neural planning baselines, while remaining competitive with classical sampling-based planners at significantly lower planning times. Ablation studies further validate the contributions of both the learned representation and the proposed planning strategy. Finally, closed-loop pushing experiments under execution deviations and external disturbances demonstrate that fast replanning enables reactive and adaptive task execution.

0.19

Robotic Spatiotemporal Editing: A Unified Interface for Controllable Action Generation via Masked Action Reconstruction

cs.RO ·2026-10-16 ·Yanbang Li, Bu He, Yi Tian, Yu-Chiao Huang, Chenyu Zhang +4 more
Abstract

Generative policies can synthesize robot behaviors, yet controlling and editing them through diverse spatiotemporal references remains challenging. In the real world, human intent requires robot behaviors that satisfy constraints on where, when, and how actions are executed. Our key insight is that spatiotemporal control references, such as keypoints, flows, orientations, and keyposes, can be viewed as partial specifications of the same temporal Cartesian action space. We introduce the MAP (Masked Action Space) interface to unify these action references and formulate controllable manipulation as an action inpainting problem. Our Inpainting Policy (IP) reconstructs missing action dimensions of the end-effector, combining progress-guided alignment with multi-mask training supporting unseen action-constraint combinations. We further adopt a training-free RePaint variant that iteratively inpaints the action during denoising to reduce control error. On ManiSkill and LIBERO-10, our policy achieves average task success rates of 92.67% and 64.67%, respectively.Real-world experiments demonstrate spatiotemporal control with average positional and rotational errors of 44.3 mm and 7°, respectively, showing that our policy can use diverse references to guide executable robot behaviors.

0.19

AWM: An All-in-One Wireless Foundation Model for Multi-User Channel Acquisition and Localization Without Downstream Tuning

eess.SP ·2026-10-16 ·Zirui Chen, Ziqing Xing, Zhaoyang Zhang, Hongning Ruan, Zili Wang +1 more
Abstract

Wireless foundation models (WFMs) seek to unify diverse wireless tasks and scenarios, yet existing methods still rely on task- or scenario-specific components or deployment-time fine-tuning. This paper presents an all-in-one wireless model (AWM), a WFM that unifies multi-user channel acquisition and localization, despite their distinct data modalities and functional objectives. We reveal that user location and channel state information (CSI) at different frequencies can be viewed as attribute-specific measurements of the same wireless link. By introducing in-scenario same-attribute reference pairs as prompts, AWM converts heterogeneous attributes into alignable relativity representations, formulating intra-band channel interpolation, inter-band channel prediction, and wireless localization as unified wireless-state transformations. Further, a multi-task MateFormer is developed to perform source-attribute relativity extraction, cross-attribute relativity alignment, and target-attribute synthesis within a unified architecture, avoiding task-specific output heads or downstream fine-tuning. Comprehensive experiments demonstrate the effectiveness of model unification, including substantial gains from cross-task knowledge sharing, seamless generalization to unseen scenarios, and an efficient closed-loop service for mobile users.

0.19

Sensing-Driven Secure Transmission with Mobile Eavesdropper Tracking in RIS-Assisted ISAC

eess.SP ·2026-10-16 ·Jie Li, Ning Wang, Zhendong Li, Ze Li, Lingfeng Shen +2 more
theory
Abstract

The mobility of potential eavesdroppers (Eves) poses a critical challenge to physical-layer security (PLS) in integrated sensing and communication (ISAC) systems, since rapidly varying Eve positions lead to time-varying eavesdropping channels and outdated security decisions. To address this challenge, this paper investigates sensing-driven secure transmission for reconfigurable intelligent surface (RIS)-assisted ISAC systems with a maneuvering Eve. To this end, by exploiting the dual functionality of the ISAC waveform, a multi-view angle estimation method jointly exploiting the base station (BS) and RIS is developed based on MUSIC and maximum likelihood estimation (MLE), enabling three-dimensional (3D) Eve localization, whose accuracy is characterized by the Cramer-Rao lower bound (CRLB). Subsequently, an interacting multiple model extended Kalman filter (IMM-EKF) with out-of-sequence measurement (OOSM) compensation is employed to track and predict the maneuvering Eve. Based on the predicted trajectory, the time-varying eavesdropping channel is updated, and the transmit beamforming and RIS phase shifts are jointly optimized subject to a sensing CRLB constraint. Furthermore, an alternating optimization (AO) algorithm integrating fractional programming, semidefinite relaxation (SDR), and Riemannian manifold optimization is developed to solve the resulting non-convex problem. Finally, numerical results demonstrate the effectiveness of the proposed approach in accurately tracking maneuvering Eves and enhancing secure transmission in dynamic ISAC environments.

0.19

Foundation Models for Autonomous Space Robotics: Architectures, Physical Adaptation, and Validation

cs.RO ·2026-10-16 ·Zizhe Wang, Jia Deng, Xin Lu, Zehao Jia, Yi Huang +2 more
Abstract · 23 pages, 7 figures. Critical review. Ancillary files include source inventory and evidence documentation

Space robots must interpret tasks and act despite unfamiliar dynamics, difficult sensing conditions, and limited opportunities for human intervention. Foundation models offer reusable perceptual and behavioral knowledge, but their value depends on how that knowledge reaches physical execution. This critical review follows space-specific requirements through perception, prediction, planning, and action to deployment and validation, emphasizing orbital manipulation and proximity operations. Terrestrial comparisons establish benefits of particular pretraining stages and model interfaces; their transfer to space dynamics and constrained onboard operation remains insufficiently tested. Event cameras offer timely motion observations across strong illumination contrasts, while spiking neural networks provide possible routes to sparse local computation. We examine how sensor degradation, state estimation and computing choices affect feedback and recovery. Higher action throughput need not yield fresher feedback; lower processor power can also coexist with higher energy per inference. We organize validation as an iterative sim-to-ground-to-space pathway, connecting physical tests with telemetry-based diagnosis and closed-loop revalidation. A satellite-capture example links observations to approach, contact and recovery, while a corresponding protocol separates generalization from adaptation with limited target-domain interaction. These comparisons identify what models already contribute and the physical evidence needed to support autonomous space operation.

0.19

FLoTS-SLAM: Optical Flow-Derived Uncertainty for Monocular Dynamic SLAM with Two-Stage Optimization

cs.RO ·2026-10-16 ·Haorui Du, Lipu Zhou
code/data
Abstract

Monocular simultaneous localization and mapping (SLAM) remains challenging in dynamic environments. Real-time dynamic object identification in such environments is a critical problem. Recent approaches estimate per-pixel uncertainty from rendering residuals or multiview feature inconsistency to identify dynamic regions. These methods lack direct motion evidence, and their accuracy may degrade when reconstruction is inaccurate or features are ambiguous. In addition, such uncertainty merely downweights pixels in bundle adjustment (BA), so inaccurate uncertainty estimates can still impair camera pose estimation and 3D reconstruction. Accordingly, we present a monocular RGB SLAM system for dynamic scenes. Specifically, a motion uncertainty estimation method is proposed to derive uncertainty from the residual between optical flow and camera motion induced flow, directly reflecting object motion. The estimation varies with frame alignment accuracy, reducing false detections and yielding accurate pixel-level masks of dynamic regions. To make BA more robust, we estimate camera poses from high confidence pixels. These poses are then used to reproject dynamic pixels during joint refinement of poses and depth, reducing the influence of object motion on both pose and depth updates. Moreover, we provide a quantitative evaluation of dynamic masks and analyze how these masks affect the downstream tracking and reconstruction of dynamic objects. Our approach improves the mean dynamic mask IoU by 35\% over a state-of-the-art baseline on DAVIS, and reduces the average ATE by 59\% on Bonn and 19\% on TUM relative to the SLAM backbone used in our system. The source code will be made publicly available.

0.19

FLASH-CBF-RRT: Fast Lipschitz-Adaptive Safe Horizon CBF Steering for Certified Motion Planning

cs.RO ·2026-10-16 ·Mani Amani, Hesam Mojtahedi, Reza Akhavian, Nikolay Atanasov, Ki Myung Brian Lee
Abstract

Fast motion planning requires both sample-efficient exploration and computationally efficient local steering. Conventional straight-line steering is inexpensive but wastes samples by rejecting colliding extensions. Quadratic program controllers with control barrier function (CBF) constraints can be used to achieve effective local steering but can increase the planning time due to the expensive sequential evaluations. We develop a method to accelerate CBF steering by computing a time horizon within which CBF-filtered velocity may be applied constantly using closed-form kinematic bounds on collision-clearance. Our method, FLASH-CBF-RRT, significantly reduces the number of evaluations compared to prior CBF-RRT approaches because the planner can take adaptive steering step sizes that are shorter near obstacles and longer in open space. We implement this steering procedure as an extension of RRT-Connect in OMPL. Across 689 MotionBenchMaker problems, FLASH-CBF-RRT matches RRT-Connect's 100% success rate while reducing median planning time by 3.82x, and is 3.88x faster than fixed-step CBF-RRT. Hardware experiments on a 17-DoF mobile manipulator demonstrate 2.2x faster planning and 4.2x fewer evaluations. Videos and results are available at: https://mamani828.github.io/FLASH-CBF-RRT/.

0.18

Unifying Embodied Foundation Models via a Model-to-Policy Formulation

cs.RO cs.AIcs.LG ·2026-10-16 ·Xinyi Hong, Mingyang Ren
Abstract

Vision-language-action (VLA) models and world-action models (WAMs) have emerged as promising approaches to robot learning and control. These models integrate multimodal perception and action generation, but differ in their architectures, output representations, and learning objectives. In this paper, we present a unified, architecture-agnostic probabilistic formulation for embodied foundation models based on stochastic kernels. Our formulation has two key components: (i) we characterize embodied foundation models through conditional output distributions and action-realization kernels, which together induce task-conditioned robot policies; and (ii) we connect pretraining and post-training to policy deployment, accommodating model-specific pretraining objectives and offline or online post-training strategies. We instantiate the formulation for OpenVLA and DreamZero, covering autoregressive action-token prediction and joint video-action generation.

0.18

CoEvolve: Self-Evolving Code Policies from Skills to Strategies for Humanoid Robots

cs.RO ·2026-10-16 ·Zicheng Liu, Haochen Wu, Jiawei He, Suyun Zhao, Huan Yang +2 more
Abstract

Recent advances in Vision-Language-Action (VLA) models and World Action Models (WAMs) have expanded the manipulation and whole body capabilities of individual robots. Yet a single robot remains limited by its embodiment, and many real world tasks require capabilities that can only emerge through cooperation. Existing multi robot frameworks largely emphasize task decomposition, allocation, and replanning, while providing limited support for adapting executable coordination when physical execution deviates from the intended behavior. We present CoEvolve, a framework that represents team coordination as executable code over fixed robot motor skills. CoEvolve introduces Verifier Guided Coordination Evolution, where a multimodal foundation model instantiates coordination strategies through motion primitives, semantic objectives, temporary robot roles, and bounded execution parameters. Physical verification evaluates each executed transition, rejected behaviors trigger structured coordination revision from an accepted checkpoint, and verified experience is retained to guide subsequent decisions. We evaluate CoEvolve with two simulated Unitree G1 humanoids transporting a freely supported rigid payload through constrained corridors. Across 15 base task runs with three multimodal backbone families, CoEvolve completes 8 tasks, with every backbone achieving end to end success. Additional evaluations yield verified placement on two additional corridor layouts and after both same-backbone and cross-backbone experience initialization, while exposing persistent limitations in motion precision and contact stability. These results support executable coordination as a practical interface for adapting long horizon multi robot collaboration under shared physical constraints.

0.18

Foundation-Model-Driven 3D Scene Graphs for Mobile Manipulation

cs.RO ·2026-10-16 ·Dominic Maggio, Lorenzo Shaikewitz, Aryan Naveen, Kris Hauser, Luca Carlone
code/data
Abstract

Geometric and Semantic Foundation models are becoming increasingly popular tools for generating spatial maps for robots. Recent work (e.g., FOUND-IT) has leveraged these tools to construct real-time, task-driven, and open-set 3D scene graphs from a single uncalibrated camera. However, while classical robot mapping techniques have undergone extensive system demonstrations, limited work has been done to develop both the fundamental and the systems wide advancements needed to integrate these new types of robot spatial memory into a full, useable autonomy pipeline. In this paper, we bridge this gap, and in doing so we also evaluate the effectiveness and limitations of this type of 3D memory. Towards this end, we integrate FOUND-IT into a mobile manipulator platform and demonstrate it is well-suited to support multiple objectives such as object change detection, support for multiple mapping sessions, determination of navigable paths, and directly support object grasping for the task of scene re-arrangement. Furthermore, we provide nuanced insights into the effectiveness and limitations of a monocular, foundation-model-driven 3D scene graph construction system using observations from rigorous on-robot experiments. Code is available at https://github. com/MIT-SPARK/FOUND-IT.

0.17

How exposed is work to robotics? Occupational, industry, and geographic measures

cs.RO cs.CYecon.GN ·2026-10-16 ·Sebastian Sartor, Robert Seamans, Neil Thompson
theory
Abstract

Exposure indices that map technical capabilities onto occupational abilities have become a standard tool for anticipating how AI may reshape work, but existing measures either cover software AI or proxy robotics through patents and installation data that reflect the industrial robots of past decades. We construct the first abilities-based exposure measures for robotics and physical AI, and the first to span occupations, industries, and geographies. Following the framework of Felten et al. (2018, 2021), we build a robotics capability taxonomy of 16 capabilities in eight domains from 78 academic, industry, and standards sources, use a large language model to score the relatedness of each capability to the 52 ONET work abilities, and aggregate to Robotics Occupational Exposure (ROE) for 774 occupations, Robotics Industry Exposure (RIE) for 250 NAICS industries, and county-level Robotics Geographic Exposure (RGE). We find that robotics exposure concentrates in physically intensive work such as construction, farming, transportation, and production, and in industries such as logging, mining, trucking, and warehousing. Unlike generative-AI exposure, ROE is higher in occupations with lower wages, lower education requirements, and less creative-ability content. Physical collaboration and assistance, terrain-based locomotion, grasping and handling, and task and functional reasoning move the exposure ranking most, yet no single capability carries the index. Because the measures capture technical relatedness rather than readiness, cost, or adoption, they should be read as an upper bound on where physical automation could affect work.

0.17

Hardware-as-Regularization: Enhancing Data Consistency in Imitation Learning via an Adjustable Passive Compliant Leader Arm

cs.RO ·2026-10-16 ·Hideyuki Ichiwara, Yushi Kaida, Hirokazu Ishida
Abstract

Low-cost passive leader arms enable scalable data collection for imitation learning, but their lack of restoring forces leaves redundant joints unconstrained, yielding highly inconsistent joint-space trajectories. We address this with a low-cost passive joint module (an adjustable mechanical clutch and a 3D-printed spiral spring) that pulls each joint toward a user-defined neutral posture. This passive restoring force acts as a physical prior equivalent to joint-space $L_2$ regularization (``Hardware-as-Regularization''), improving kinematic consistency with shorter completion time than an unregularized arm and preserved movement smoothness. Experiments show substantial variance reduction over a naive passive arm, trajectory consistency on par with an active virtual spring of the same actuator class, and substantially higher downstream success rates (e.g., 73.3% vs. 36.7% on Block Stacking); a stiffness sweep shows consistency degrading monotonically as stiffness decreases, confirming that mechanical stiffness behaves like a regularization weight.

0.16

SkyAgent: Revisable Target Memory for Continuous Aerial Search

cs.RO ·2026-10-16 ·Kaibo Yang, Xingyu Wu, Enze Zhu, Yajie Chai, Xuyang Li +4 more
Abstract

Continuous aerial search builds target hypotheses from sparse, viewpoint-dependent observations. When later evidence contradicts a stored hypothesis, however, the candidate can remain available to the planner and continue triggering approach or verification actions. We call this trajectory-level failure false candidate persistence (FCP). We equip SkyAgent with Revisable Target Memory (RTM), which associates evidence with persistent candidates, discounts correlated views, and revises each candidate through Tentative, Confirmed, and Rejected states. The key interface binds every generated waypoint to its source candidate: rejection therefore invalidates stale goals, while conservative reactivation protects temporarily occluded targets. Across four splits from the CityAVOS, UAV-ON, and OpenUAV benchmarks, our SkyAgent equipped with RTM outperforms all competitors, improving SR by 2.90--6.31 points while substantially reducing Behavior-FCP and post-contradiction travel with a low false-rejection rate. Across 35 matched real-world DJI scenarios per method, RTM consistently improves task success and behavioral reliability, further validating its state-to-action correction under physical deployment conditions.

0.16

Data Center Load Accommodation under QoS and Checkpoint-Induced Frequency Security Constraints

eess.SY ·2026-10-16 ·Ruolan Cui, Ziang Liu, Chenxi Hu, Francis Yunhe Hou
benchmark
Abstract · 10 pages

The rapid growth of artificial intelligence workloads is accelerating the expansion and grid interconnection of data centers (DCs). Existing data center load accommodation (DCLA) assessments primarily focus on grid and resource adequacy and often imply that greater workload flexibility always increases accommodable capacity. However, workload flexibility has a dual effect. Workload deferral and temporal shifting can alleviate network congestion, whereas synchronized checkpointing of training workloads can cause large and rapid load reductions, threatening frequency stability. Therefore, DCLA cannot be determined solely by grid adequacy. This paper develops an optimal DCLA method that jointly accounts for workload quality of service (QoS), checkpoint-induced frequency security (FS), and grid adequacy. First, a QoS-aware model characterizes workload-dependent flexibility while ensuring service feasibility. Second, checkpoint-induced FS constraints are developed for periodic and event-triggered power variations, with key frequency-security limits reformulated in algebraic form. Third, a two-stage lexicographic model is developed to maximize DCLA and minimizes the required energy storage (ES) deployment, and then optimizes system operation under the given DC and ES siting and sizing decisions. Case studies show that the proposed method avoids a 40.3% overestimation of DCLA compared with the grid-adequacy benchmark. QoS is the primary limiting factor, while checkpoint-induced frequency-security constraints become more restrictive as training workloads increase. These findings highlight that reliable DCLA should account for both its grid-support value and the associated QoS and FS risks.

0.16

AWAM: Active Trajectory-Conditioned View Imagination for Robust Robot Manipulation

cs.RO cs.AI ·2026-10-16 ·Jiawei Geng, Zhenyang Liu, Yuzheng Wu, Yanwei Fu
Abstract

Rapid progress in embodied intelligence has enabled increasingly capable robots to follow language instructions and manipulate diverse objects. However, strong results in structured environments do not automatically transfer to open-world conditions, where clutter, occlusion, and changing viewpoints can obscure the spatial relationships required for action. To address this perceptual gap, a virtual active-perception paradigm is introduced. AWAM, an Active World-Action Model actively plans where to observe and imagines what will be seen along the planned trajectory before acting. Task-related virtual camera motion is predicted and used as a structured query into a geometry-conditioned world model, producing a reusable spatial prior for manipulation, which is later compressed into a fixed memory and fused with live observations through cross-attention for diffusion-based action prediction. AWAM achieves competitive success rates with 75.9% on LIBERO-Occ, 62.4% on LIBERO-Plus, 76.5% on RLBench-OG Occlusion1, and 94.4% in real-robot trials. These results support improved robustness to the evaluated occlusions and distribution shifts, addressing part of the gap between structured manipulation and less controlled environments.

0.16

DefFlow: Goal-Conditioned Deformable Object Manipulation using 3D Point Flows

cs.RO ·2026-10-16 ·Gaurav Singh, Sergio Orozco, Ryan Duong, Mete Tuluhan Akbulut, George Konidaris +1 more
Abstract

Manipulating deformable objects toward desired 3D configurations is a fundamental capability for robots operating in unstructured physical environments. However, as the range of possible initial and target configurations grows, goal-conditioned deformable manipulation requires a robot to infer increasingly diverse manipulation behaviors in addition to precise continuous actions. We introduce DefFlow, a framework for goal-conditioned manipulation of deformable objects toward a broad range of 3D configurations. DefFlow learns from demonstrations by jointly predicting the goal-conditioned 3D point flow of the robot and deformable object. We show that this formulation better captures both the manipulation decisions, such as when and where to grasp, release, or re-grasp, and the precise continuous motions required for goal reaching, rather than predicting only the robot flow. Across rope, granular-media, and cloth manipulation settings with diverse initial and target configurations, DefFlow substantially outperforms goal-conditioned action-prediction and robot-flow-only baselines on success rates and goal-distance metrics. We further demonstrate deployment of our framework on a real robot using real point-cloud observations and unseen goal configurations.

0.15

Evaluating Lightweight Contact Feedback in Dual-Operator Collaborative Humanoid Teleoperation

cs.RO cs.HC ·2026-10-16 ·Austin Graves, Chuizheng Kong, Yunho Cho, Yuanshao Yang, Advaith Nidumukkala +2 more
Abstract · 8 pages, 4 figures, 2 tables. Project website: https://xrt-teleop.github.io

Dual-operator humanoid teleoperation enables collaborative manipulation through two separate robots, but operators lack direct awareness of robot-object contact that is normally available during physical interaction. To address this, we present eXtended Reality Teleoperation (XRT), an extended reality teleoperation framework that returns a robot's fingertip contact through lightweight vibrotactile and/or auditory feedback, and a synchronized dual-humanoid testbed built on it for two operators. Sixteen dyads performed a collaborative crate lift-and-flip under vision only and the three feedback conditions in a counterbalanced within-subject design with verified feedback-channel delivery. Three results follow. First, operators rated our finger-localized vibrotactile display as more effective than our open-ear sonification: all 32 ranked it above the sonification, 18 ranked it most effective, and its cues were rated far easier to perceive and interpret than the auditory cues. Second, completion time, failures, workload, discomfort, and partner awareness showed no detected condition differences, with wide completion-time intervals and smaller meaningful effects unresolved. Third, performance changed substantially with practice, especially on the harder flip-over task: completion time improved across blocks, and attempt-to-attempt variation within a dyad (SD x1.51) was larger than the largest condition contrast (x1.16). Exploratory analyses detected no difference in immediate grip or motion responses between feedback conditions. For novice collaborative teleoperation, our fingertip vibrotactile cues were rated more effective and easier to interpret than our auditory cues, while their effect on task performance remains unresolved.

0.15

LWMa: Distributed Deployment of an Angle-Delay-Domain Wireless Foundation Model for Multiple Tasks

eess.SP ·2026-10-16 ·Zhizhen Ren, Xudong Zhang, Jingbo Tan, Jintao Wang
Abstract

The growing diversity of wireless services in sixth-generation (6G) wireless communication systems calls for reusable channel representations that support multiple tasks across different propagation environments. However, deploying such representations between user equipment and the base station in frequency-division duplexing systems faces uplink feedback constraints. This paper proposes a distributed framework based on an angle-delay-domain large wireless model (LWMa) for multiple tasks. LWMa exploits angle-delay structure and delay truncation to improve feature extraction and reduce computational cost. A scenario-specific network inspired by Perceiver IO (input-output) compresses embeddings for channel reconstruction tasks, while other tasks use a low-dimensional compact global embedding. Embedding reconstruction training followed by channel-quality-guided fine-tuning preserves information useful for downstream processing. Simulations demonstrate that the proposed method supports effective channel reconstruction and multiple tasks under limited feedback with reduced deployment complexity. Further evaluations show that the recovered embeddings generalize to additional tasks and support split inference.

0.15

Ego2Contact: Predicting Touch from Egocentric Human Video for Contact-Rich Robot Manipulation

cs.RO ·2026-10-16 ·Qianqian Yang, Dayou Li, Bangya Liu, Zihao Zhu, Nuo Chen +9 more
Abstract · 8 pages, 6 figures

Contact locations and interaction forces are critical for contact-rich robot manipulation but are often ambiguous from vision alone. Measuring them on robots requires specialized tactile hardware and platform-specific data collection, which keeps robot tactile data scarce. Human egocentric video is far easier to scale but lacks touch. Recent predictors trained on egocentric recordings paired with tactile-glove measurements can predict this missing signal from video alone. These predictors, however, remain confined to human hands, which differ from robot hands in appearance and tactile layout. We present Ego2Contact, which learns a visual-tactile prior from human interactions and transfers it to robot hands. Its predictor fuses temporal scene context with hand-local visual features to infer bimanual tactile fields from the RGB video. Anatomical region correspondences align varying robot tactile layouts with the predictor's canonical grid, turning robot measurements into fine-tuning targets and predictions into policy inputs. Ego2Contact outperforms prior tactile predictors on human data, including unseen objects, and on robot data after adaptation. The adapted predictor provides tactile inputs for both simulated reinforcement learning and real-robot policy post-training. In simulation, policies using predicted touch outperform non-tactile baselines and recover part of the benefit of ground-truth tactile sensing. These results show that human-learned tactile priors can supply touch when direct tactile sensing is limited.

0.15

Beyond Action Prediction: Learning Search-Free Planning in Latent Space

cs.RO cs.AI ·2026-10-16 ·Guanchen Du, Jinhao Du, Yanpeng Sun, Shan Zhang, Na Zhao +1 more
Abstract

Latent world models learn action-conditioned state transitions in a compact representation space, providing a predictive model for visual planning and control. However, planning with these models typically requires costly test-time optimization over action sequences. We propose to replace this online search with a goal-conditioned inverse model that directly predicts actions from current and goal observations. A key challenge is that action prediction alone neither ensures that goal representations encode action-relevant information nor that predicted actions are consistent with the learned latent prediction model. We introduce Hindsight Inverse JEPA Planning (HIJP), a search-free framework that addresses both aspects during training. HIJP relabels future observations along offline trajectories as goals, providing action supervision across multiple temporal horizons. By retaining gradients through the goal representations, this supervision directly shapes the visual representation. HIJP further feeds predicted actions into the latent predictor model and aligns the resulting next-state representations with encoded successor observations, enforcing action-transition consistency. At inference time, the inverse dynamics model and latent predictor models are applied alternately to generate multi-step action sequences without online search. Across four visual control tasks, HIJP achieves 86.8% overall success and a 1.9-16.5x planning speedup over search-based planning.

0.15

A Smaller Subspace is All You Need: Efficient ISAC Beamforming via Dimensionality Reduction

eess.SP cs.IT ·2026-10-16 ·Ruiqi Liu, Yuanshuo Gang, Tianqi Mao, Qingqing Wu
Abstract · Submitted to an IEEE conference

Integrated sensing and communication (ISAC) has been recognized as a critical technology for the sixth generation (6G) wireless networks. Under monostatic sensing mode, achieving a balance between communication and sensing performance via optimal beamforming is a critical research problem. In multiple input multiple output (MIMO) systems, such beamforming problems are of substantial computational complexity due to the large number of antennas. Existing methods exploit exact subspaces of lower dimensions to speed up the beamforming process where the complexity is proportional to the number of communication users and sensing targets instead of antennas. In this paper, it is demonstrated that the beamforming problem can be solved in an even lower-dimensional subspace with only a small sensing performance loss. Numerical results compare the sensing performance and computational complexity of ISAC beamforming problem in the exact subspace and smaller subspaces.

0.14

Converter-Aware Coordination of Computing Flexibility in Behind-the-Meter Data Centers

eess.SY ·2026-10-16 ·Ruolan Cui, Ruizhang Yang, Chenxi Hu, Franxis Yunhe Hou
Abstract

Data centers are emerging as large electricity consumers whose demand can be reshaped through workload scheduling. However, computing flexibility in behind-the-meter facilities does not automatically translate into electrically realizable power flexibility. Ignoring converter and network operating limits may cause converter overloads, protection trips, and interruptions to power supply and computing services. This paper proposes a converter-aware intraday scheduling methodology. First, a task-level mixed-integer linear formulation preserves synchronous training configurations and completion requirements, while Minkowski-sum aggregation and Shapley--Folkman convexification yield a linear inference-pool representation without unit-level binary variables. The latter admits a conditional relative relaxation-gap bound that decreases inversely with pool size. Second, a four-node electrical model links computing demand, on-site injections, and grid exchange to converter and network limits through lifted voltage variables and second-order cone relaxation. Finally, a mixed-integer model coordinates workloads, photovoltaic generation, energy storage, and grid purchases. Case studies show that electrical limits reshape scheduling and renewable utilization, while on-site resources reduce grid dependence. Under the studied conditions, fixed training cannot support full inference service within voltage limits, whereas coordinated scheduling achieves negligible inference shortage and passes verification against the original alternating-current model. These findings demonstrate the role of computing--electrical coordination in realizing the operational value of workload flexibility.

0.14

Filter-Augmented Reinforcement Learning for Battery Fast Charging with State of Health and Plating-Aware Reward Design

eess.SY ·2026-10-16 ·Meng Yuan, Yang Li, Changfu Zou
Abstract

Fast charging of lithium-ion batteries is crucial for electric vehicle adoption, yet it increases safety risks and accelerates battery ageing. To address this, we propose a filter-augmented reinforcement learning (RL) framework for lifelong health-aware fast charging, based on the twin-delayed deep deterministic policy gradient (TD3) algorithm. The RL agent is guided by a reward function that balances charging speed and battery longevity through the incorporation of state of health (SoH) information and an SoH-dependent voltage penalty term derived from a constant--current constant--overpotential charging strategy. Since onboard SoH estimates are inevitably corrupted by noise, two filtering strategies are integrated: an exponential moving average (EMA) filter when noise statistics are unknown, and a Kalman filter when they are available, with the filtered estimate embedded in both the agent's state and reward computation. The RL policy is trained using data from high-fidelity life-cycle simulations. The evaluation uses a published electrochemical and degradation parameterisation for a commercial LG M50 cell. Results show that the proposed method achieves up to 21.4\% more equivalent full cycles (EFCs) than conventional constant current--constant voltage (CC--CV) charging under 2\% SoH estimation noise, and maintains this lifetime advantage even under 5\% noise, while sustaining comparable charging speed.

0.14

No Eyes, No Problem: What Does Touch Contribute to Blind Baoding Ball Manipulation?

cs.RO ·2026-10-16 ·Ayush Deshmukh, Nalin Agarwal, Elle Miller, Sethu Vijayakumar
benchmark
Abstract · 9 Pages, 8 Figures

Humans manipulate objects within their hands through touch alone with remarkable ease, but for robots, achieving sustained real-world manipulation without vision remains an open challenge. Although reinforcement learning (RL) can generate sophisticated blind manipulation policies in simulation, transferring them to a real dexterous hand remains hindered by continuously changing contacts, motion-induced tactile information drift, and the gap between simulated and physical dynamics. We study this problem through Baoding ball rotation on the Shadow Dexterous Hand Lite, using only proprioception and binary tactile feedback as observations. We achieve successful zero-shot transfer through a combination of a sensor-aligned, coupling-aware embodiment in simulation, tactile calibration, and two novel episodic domain randomisations: slew-rate action limits and STuck-AT Taxels. On the physical hand, the policy achieves sustained blind Baoding rotation, completing $\textbf{225 consecutive $\mathbf{180^\circ}$ rotations (112.5 full rotations) at $\sim0.36 \;\text{rot/s}$}$ in a single continuous trial without any vision or object-state information. Furthermore, we evaluate the contribution of tactile sensing relative to proprioception alone and identify the hand regions where tactile information is most important. This is the strongest reported hardware result for Baoding ball manipulation in both rotation count and speed, irrespective of the sensing modality used. By effectively bridging the sim-to-real gap, our framework sets a new benchmark for sustained tactile-driven dexterous manipulation on physical hardware.

0.13

ECG-SDI: Single-lead Electrocardiogram Enables Continuous Sleep Depth Index Profiling and Scalable Digital Biomarkers for Sleep Health

eess.SP cs.LG ·2026-10-16 ·Donglin Xie, Qingshuo Zhao, Songchi Zhou, Haoyu Wang, Jingyu Wang +3 more
Abstract

Conventional sleep staging provides a coarse-grained representation that obscures within-stage depth variation. We developed ECG-SDI, a single-lead electrocardiography (ECG) framework for continuous sleep-depth profiling and digital biomarker extraction. Using 12,392 overnight polysomnographic recordings, we trained and validated a two-stage deep learning model supervised by conventional sleep-stage labels. Correlations with ordered stages excluding rapid eye movement (REM) sleep were 0.828 in the pooled internal test set and 0.848 in the external Cleveland Family Study; corresponding REM classification AUROCs were 0.954 and 0.977. Overnight trajectories captured within-stage variation. Biomarker-based phenotyping identified a disturbed subtype associated with poorer subjective sleep quality and higher all-cause mortality in the Sleep Heart Health Study (adjusted hazard ratio, 1.20; 95% CI, 1.05-1.38). Among 507 participants with paired recordings before and after continuous positive airway pressure (CPAP), normalized sleep-depth area increased by 6.1% and shallow-sleep burden below 0.3 decreased by 16.0%. Exploratory analyses linked reductions in approximate entropy and depth variability to reduced oxygen desaturation and increased sleep efficiency, respectively (Spearman rho = 0.24 and 0.12); both associations remained positive after covariate adjustment. Mean ECG-SDI within N2 increased by 0.017 (95% CI, 0.004-0.030). A wearable proof-of-concept illustrated the acquisition-to-report workflow. These findings support ECG-based sleep phenotyping and further evaluation of CPAP response monitoring at home.

0.13

TrajEdit: Real-Time Flow-Matching Trajectory Adaptation via FlowEdit and CBF Safety Filter

cs.RO ·2026-10-16 ·Wondmgezahu Teshome, Arya Honarpisheh, Octavia Camps, Mario Sznaier
Abstract · 8 pages, 3 figures, 4 tables

Safe, real-time motion planning is especially difficult when a dynamic obstacle actively pursues the ego agent, requiring rapid adaptation as the environment changes. We present TrajEdit, a hierarchical trajectory planner for pursuit-evasion that combines flow matching, trajectory editing, and safety filtering. Rather than regenerating trajectories from scratch at every step, TrajEdit follows the current plan under nominal conditions and adapts its remaining portion through FlowEdit as the pursuer approaches, and falls back to a control barrier function safety filter if the adversarial agent is within a safety threshold. Our use of FlowEdit for trajectory editing reduces computation time while maintaining a comparable success rate. We evaluate TrajEdit in 2D pursuit-evasion with static obstacles and an adversarial pursuer, demonstrating high success rates with low capture and collision rates, while achieving computational time savings over full replanning baselines.

0.13

Beyond Human-Derived Motion: Exploring Robot-Specific Execution for Construction Humanoids

cs.RO ·2026-10-16 ·MinJun Chang, Taehyung Kim, Yong K. Cho, Francis Baek
Abstract

Construction productivity has remained a per- sistent challenge, motivating growing interest in humanoid robots that can operate in workspaces and with tools designed for human workers. Realizing this potential requires effective motion control, for which human-derived motion transfer has emerged as a promising approach by capturing and retargeting human demonstrations to humanoid robots. While promising, however, human-derived motion transfer may not fully define how a humanoid should ultimately execute the transferred motion. Once task-relevant motion information has been con- veyed, the humanoid may also exploit its own embodiment through motions that depart from the human demonstration. This study examines whether such Robot-Specific Motion offers an advantage in motion execution efficiency over Human- Derived Motion. On an adult-sized Unitree H1-2, Human- Derived Motion was generated through vision-based capture and retargeting, whereas Robot-Specific Motion used inverse kinematics and minimum-jerk transfers to reach the same targets while departing from the demonstrated posture, as verified using Rapid Upper Limb Assessment (RULA). Across a pick-and-stack sequence representing construction material handling, with four matched motion targets and three trials per condition, Robot-Specific Motion achieved 12.87 s versus 25.03 s in execution time, 0.92 m versus 4.33 m in trajectory length, and 58.7 J versus 247.3 J in two-arm mechanical work. These findings suggest that allowing humanoids to depart from demonstrated human motion may offer additional opportunities for more efficient motion execution, while broader construction tasks are needed to determine whether such differences trans- late into productivity gains.

0.09

Reinforcement Learning Based Multi-Terrain Adaptive Whole-Body Control for a Quadruped Manipulator

cs.RO ·2026-10-16 ·Yanzhou Jin, Fei Suo, Chang Liu, Yasuhisa Hasegawa, Liming Chen +3 more
Abstract

Legged robots equipped with manipulators offer promising capabilities for loco-manipulation tasks in unstructured environments. However, achieving task-consistent end-effector behavior during locomotion remains challenging due to terrain-induced disturbances and the strong coupling between locomotion and manipulation, particularly for small-scale quadruped platforms. A reinforcement-learning based whole-body control framework is proposed for quadruped loco-manipulation on complex terrain. The proposed framework combines a hybrid task representation with task-prioritized whole-body learning to coordinate locomotion and manipulation while preserving task-consistent end-effector behavior. A lightweight quadruped manipulator platform is further developed to validate the proposed framework on a small-scale quadruped system. Real-world experiments demonstrate successful deployment on complex terrains such as slopes up to 30 degrees, stairs, and deformable terrain. These results demonstrate the feasibility of deploying learned whole-body control policy for quadruped loco-manipulation in real-world environments and highlight the importance of task-consistent whole-body coordination for terrain-adaptive loco-manipulation.

0.09

Enhancing Visual Task-Redirection Attacks on Vision-Language-Action Models via Closed-Loop Optimization

cs.RO ·2026-10-16 ·Chunjin Ye, Guanying Chen, Jianjie Huang, Jianhao Chen, Quanlin Chen +4 more
Abstract

Vision-language-action (VLA) models enable language-guided robot manipulation, but remain vulnerable to visual attacks that undermine user control over robot behavior. Prior studies optimize visual perturbations using fixed target supervision. However, the perturbation can change earlier actions and, in turn, the physical states reached later. As the attacked trajectory evolves, this supervision may no longer match the behavior needed at the states actually reached. Accurately matching these targets can therefore fail to advance the attacker-specified task. Even with suitable supervision, reducing average matching error does not ensure sustained task progress. The robot may still fail at a critical stage, and further updates can disrupt previously successful behavior. To address these limitations, we propose TRACO, a framework that enhances existing targeted visual attacks through closed-loop optimization. Its Execution-Driven Supervision rebuilds supervision for the states reached under attack to guide subsequent updates. Within this framework, Progress-Guided Selection focuses updates where progress stalls while maintaining supervision on the earlier behavior needed to reach that point. Experiments across VLA models and simulation environments show improved task-redirection success with TRACO, highlighting the security implications of closed-loop attack optimization.

0.08

People & Society 41

Immersive Focus+Context Techniques to Assist in Interpretation of High Density Forest Point Clouds

cs.HC ·2026-10-16 ·Spencer O'Keeffe, Jan Rombouts, Bruce H Thomas, Michelle Balasso, Jim O'Hehir +1 more
top venuecode/datahighly cited
Abstract

Virtual Reality (VR) is changing how we interact with complex spatial datasets. VR has the potential within forestry to help analyse remotely sensed spatial data for inventory management. However, challenges of scale and legibility of dense forest point clouds must be addressed. To do so, we present three VR Focus+Context (F+C) techniques: Wedgelight, Remote Lantern, and Ray Lantern. These techniques present point cloud data using a constrained highdetail focus within a broader abstracted context representation, manipulated using egocentric and exocentric interactions. We conducted two user studies. The first informed the design of the F+C techniques. The results from the first study indicate that VR significantly outperforms desktop displays in basic analysis of forestry point clouds, with improvements to both time and accuracy. The second study evaluated the performance of the resulting F+C techniques in a task analogous to real-world forestry operations. Our F+C techniques significantly outperformed current approaches, confirming their viability for facilitating detailed and effective forest assessments. These findings indicate that VR has significant advantages for spatial data analysis tasks in forestry, making a strong argument for its adoption in future analytical frameworks. The F+C visualisation techniques we introduced show considerable promise for analysis of high density point clouds, opening new avenues for development in immersive analytics. Supplemental materials (code, scans, results) are available at https://osf.io/hqc9u.

0.82

SwitchTube: A Proof-of-Concept System Introducing "Adaptable Commitment Interfaces" as a Tool for Digital Wellbeing

cs.HC ·2026-10-16 ·Kai Lukoff, Ulrik Lyngs, Karina Shirokova, Raveena Rao, Larry Tian +3 more
top venuehighly cited
Abstract · 22 pages. Published in Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (CHI '23)

YouTube has many features, such as homepage recommendations, that encourage users to explore its vast library of videos. However, when users visit YouTube with a specific intention, e.g., learning how to program in Python, these features to encourage exploration are often distracting. Prior work has innovated 'commitment interfaces' that restrict social media but finds that they often indiscriminately block needed content. In this paper, we describe the design, development, and evaluation of an 'adaptable commitment interface,' the SwitchTube mobile app, in which users can toggle between two interfaces when watching YouTube videos: Focus Mode (search-first) and Explore Mode (recommendations-first). In a three-week field deployment with 46 US participants, we evaluate how the ability to switch between interfaces affects user experience, finding that it provides users with a greater sense of agency, satisfaction, and goal alignment. We conclude with design implications for how adaptable commitment interfaces can support digital wellbeing.

0.74

mmPencil: Toward Writing-Style-Independent In-Air Handwriting Recognition via mmWave Radar and Large Vision-Language Model

cs.HC cs.CV ·2026-10-16 ·Yifan Guo, Zhu Wang, Qian Qin, Yangqian Lei, Qiwen Gan +4 more
publisheddatasethighly cited
Abstract · Published in ACM IMWUT, Vol. 9, No. 3, Article 81, 2025; presented at ACM UbiComp 2025. 30 pages

In-air handwriting recognition based on radio frequency (RF) signals offers a promising solution for human-computer interaction, yet existing methods are often limited by predefined writing planes, constrained writing styles, and small recognition vocabularies. To address these limitations, we propose m^2VLMs, a modality mapping architecture that bridges millimeter-wave (mmWave) radar sensing and large vision-language models (VLMs), and develop mmPencil, a 3D in-air handwriting word recognition system. mmPencil reconstructs 3D handwriting trajectories from mmWave signals using a multi-stage spatial trajectory reconstruction algorithm and maps them into visual representations through a spatial-to-visual mapping algorithm, enabling VLMs to recognize handwriting while reducing sensitivity to writing styles and environmental factors. We collect and release a 3D mmWave handwriting dataset containing 7,664 samples of 200 words from 12 users across 22 writing scenarios. Extensive experiments demonstrate that mmPencil achieves robust word recognition under variations in writing position, range, angle, speed, size, direction, and user movement. It achieves 97.60% accuracy on 200 word classes for seen users and 92.50% zero-shot accuracy on 50 word classes for unseen users, outperforming state-of-the-art baselines.

0.64

PersonaCUABench: Benchmarking Computer-Use Agents Serving Diverse Personas

cs.HC cs.AIcs.CYcs.LG ·2026-10-16 ·Haoming Li, Defu Cao, Xu Hu, Yingtao Luo, Yiyang Wang +21 more
benchmarkcode/data
Abstract · 31 pages, 13 figures, 24 tables

Computer-use agents (CUAs) are commonly evaluated against fixed task specifications, but users may omit requirements, add constraints, or require approval before consequential actions. Completing the benchmark instruction does not guarantee serving the person who asked. We introduce PersonaCUABench, which holds 74 tasks across 11 domains fixed and varies only who asks. A persona agent, playing one of 44 personas in four groups, rewrites each instruction in that user's own words and answers the CUA's questions, while the CUA sees only the dialogue, never the profile or the original instruction. Across nine backbones (3,330 runs), mean task completion falls by 36% relative to the original requests. The losses have two sources: in 59.9% of cells that lose points, the agent engages with what the user asked for and still gets it wrong, and 11.2% are stops the user requested that the checker scores as failures. Even when a persona request is fully completed, the user gives the top satisfaction rating in only 62.7% of cells, against 84.1% for the original requests. CUAs should therefore be evaluated not only on task completion but also on whether they meet the requirements and constraints users express. Code: https://github.com/MatrAIx-ai/PersonaCUABench. Project page: https://matraix.ai/research/persona-cua.html

0.64

When Ubiquitous Health Data Stops Helping: Patients' Experience of Breakdowns in the Data-Interpretation-Action Loop at the End of Life

cs.HC ·2026-10-16 ·Yibo Meng, Xinru Wang, Zhuoyan Li, Chengxi Zang, Zhuoran Lu +1 more
published
Abstract · Accepted to Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), Vol. 10, No. 4, 2026

Ubiquitous computing systems in healthcare often assume that greater data visibility improves awareness, interpretation, and action. In end-of-life care, however, this assumption warrants scrutiny. Through semi-structured interviews with 25 terminal-stage patients as primary informants and 8 clinicians as supplementary informants, we examine patients' descriptions of how they encountered, interpreted, acted on, and disengaged from state signals, trajectories, algorithmic outputs, and behavioral audits. Our analysis identifies three recurrent patterns in these descriptions. First, participants described repeated and unresolved sensemaking when visible data lacked explanatory pathways. Second, they described outputs they could understand but whose suggested actions did not fit their changing bodily capacities, care goals, or decision-making roles. Third, they described difficulty disengaging when data remained embedded across care environments. Rather than rejecting data altogether, patients called for less data exposure and a reorientation toward existential support. Based on this analysis, we situate data exposure as a central interaction problem in end-of-life ubiquitous computing and derive design implications for interpretive stability, care-context alignment, and existential support.

0.64

"Can Computing Systems Eventually Disappear?": A Long-Term Tracking Study on the Use of Diabetes Computing Systems by Patients with Diabetes

cs.HC ·2026-10-16 ·Yibo Meng, Jinyuan Zhan, Yao Lyu, Zhuoran Lu, Chengxi Zang
publishedposition
Abstract · Accepted to Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), Vol. 10, No. 4, 2026

Ubiquitous computing has long pursued Weiser's vision of technologies that "disappear" into everyday life. As wearable health systems advance, they increasingly substitute for rather than scaffold users' judgment, yet the long-term consequences remain underexplored. We report an eight-year, five-wave longitudinal interview study from 2017 to 2025 with 37 patients with diabetes, complemented by interviews with 15 clinicians, designers, and vendors. Across three technological generations including glucometers, continuous glucose monitoring, and predictive and automated insulin delivery systems, we identify a novel user state, continued use with cognitive disengagement. In this state, devices remain in use but active interpretation and questioning recede, leaving users to respond only to alarms. This state emerges from functional expansion, delegation of interpretive authority, and information fatigue, and is reinforced by clinical prioritization, information compression, and automation. It results in less frequent independent health judgment and a shift from understanding to responding. We argue that future diabetes computing systems should reduce unnecessary cognitive work while preserving users' opportunities to interpret, contextualize, and question system outputs.

0.60

"You Already Have Me--Why Do You Still Use Blued?" How Gay Dating Apps Shape Intimate Relationships Among Chinese Gay Men

cs.HC ·2026-10-16 ·Yibo Meng, Lyumanshan Ye, Bingyi Liu, Linghao Li, Huidi Lu +1 more
published
Abstract · Accepted to ACM GROUP 2027

Gay dating applications such as Blued are widely used for meeting partners, yet their role does not end once an intimate relationship is formed. When these platforms persist within an established relationship, they transform from matchmaking tools into a form of relational infrastructure that both partners must continuously manage. Building on existing CSCW and HCI research that has primarily examined how social media triggers jealousy, surveillance, or trust issues in intimate relationships, this study investigates how the persistent presence of gay dating applications shapes the dynamics of intimate relationships among Chinese gay men. Through semi-structured interviews with 23 self-identified gay men aged 19 to 55, we identify a recurring cyclical process: platform signals-such as online status, profile updates, distance indicators, and usage records-trigger suspicion and conflict; partners invest negotiation labor through intent explanation, visibility control, rule formation, or silence; and these negotiations eventually sediment into temporary governance settlements that reduce trigger frequency but remain conditionally stable. We identify four types of governance settlements: infrastructure removal, transparency governance, silence agreements, and layered/open agreements. Each settlement addresses the governance problems of platform presence, visibility distribution, and loyalty boundaries differently, yet none achieves permanent stability due to platforms' reversible entry mechanisms and continuous visibility cues. Our findings extend existing CSCW research on digital technology and intimate relationships by revealing a more complex process than previously documented: platform signals not only trigger emotional responses but prompt partners to engage in sustained relational negotiation through explanation, visibility management, rule formation, or silence.

0.47

Navigating Deception on Gay Dating Apps: A Qualitative Study of Users' Experiences, Verification Strategies, and Community Knowledge in China

cs.HC ·2026-10-16 ·Yibo Meng, Lyumanshan Ye, Bingyi Liu, Huidi Lu, Xiaolan Ding
published
Abstract · Accepted to ACM GROUP 2027

Gay dating applications have become essential platforms for sexual minority men to seek relationships, intimacy, and community, yet they also expose users to a range of deceptive interactions that remain underexplored in HCI and CSCW research. This study examines how gay male users in China experience, identify, and respond to deception on dating applications. Through semi-structured interviews with 22 participants who used platforms such as Blued, Aloha, Fanka, and Soul, we identify a typology of deceptive practices that users encounter, including relational misrepresentation for sexual access, sustained emotional deception, financial fraud, marketing behavior disguised as ordinary users, and false identity construction. We further analyze the verification and coping strategies that users develop through long-term platform use, such as judging relational intent through interactional pace, identifying financial risk signals, recognizing commercial promotion accounts, and progressively verifying identity authenticity. Finally, we show that risk recognition is not merely an individual capacity but a collaborative practice that gradually emerges through the circulation of experience within the community, the summarization of recurrent deceptive tactics, and the learning of shared rules. Our findings contribute to CSCW scholarship by demonstrating how users of a marginalized community collectively construct practical knowledge about deception, and by revealing that trust assessment on dating platforms operates as a layered, probabilistic, and socially distributed process rather than a binary judgment.

0.42

Designing for Imperfect Prompts: A Simulation-Based Evaluation of an LLM-Based Enterprise Data Assistant Using Knowledge Graphs

cs.HC cs.AIcs.DB ·2026-10-16 ·Niklas Wagner, Alexander Mädche
published
Abstract · Published in the Proceedings of the 34th European Conference on Information Systems (ECIS 2026)

Enterprise data assistants using large language models (LLMs) promise democratized data access in organizations. Yet in our real-world deployment within the CFO unit's data lake at an industry partner, we observed that business users frequently pose imperfect prompts that include domain-specific vocabulary and lack schema references or structural information. This Design Science Research study examines automatic resolution using a semantic mediation layer. We design and evaluate two design variants: schema-based mediation via semantic similarity, and knowledge graph-based mediation leveraging an automatically constructed knowledge graph from the data lake. Our simulation-based evaluation using real-world data, supported by LLM-as-a-Judge assessments and human validation, shows that knowledge graph-based mediation outperforms the schema-based approach in Precision, Recall, F1, and preserving user intent. We propose a design principle for knowledge graph-based mediation, demonstrating that structured knowledge representation bridges user intent and database structure and guides the design of enterprise data assistants across skill levels.

0.41

"You" Is Not Toxic: How LLM Labelers Rewrite the Web's Conversation Laws

cs.SI cs.CL ·2026-10-16 ·Liang Yue, Eric A. Vance
theorycode/data
Abstract · 43 pages, 4 figures, 44 tables. Labels, prompts and code: https://github.com/liyu2195/you-is-not-toxic

What the Web community believes about toxic conversations, such as that toxicity rises in longer and deeper threads, was measured mostly with Google's Perspective API, which is shutting down; large language models (LLMs) will make the next measurements. We find that LLM labelers mistake second-person address for toxicity, and that this one bias manufactures laws that human labels do not show. On 1.25 million news comments with crowd labels, Claude Haiku flags 12% of the non-toxic comments that contain no "you" and 51% of those with four or more; a single crowd rater flags 6% and 10%. Deeper replies address someone more often, so under Haiku toxicity rises with reply depth (+0.93 percentage points per level), while under the crowd's labels it slightly declines (-0.12). Rewriting "you" in the third person removes two thirds of the false rise; flagging only as many comments as the crowd does not remove it, and showing Haiku the parent comment enlarges it. All seven LLMs we test over-flag "you", and at the crowd's flag rate, without the parent, any false rise of the open ones vanishes at a fixed number of second-person words; Haiku also over-flags "you" on Reddit, Gab and Wikipedia talk pages. Agreement with human labels does not reveal the bias: we prove that a fixed Cohen's kappa leaves a law nearly unconstrained. We therefore certify the law instead of the labeler. Threadsafe uses the labeler's scores to choose which comments human raters label and to predict the others, and corrects the predictions with the human labels; linear laws are estimated without bias whatever the labeler does. In simulated audits its intervals for the laws keep nominal coverage from 250 labels on, where uniform-sample methods such as PPI++ fall short, and at 1,000 to 8,000 labels are 0.75 times as wide as those of design-based supervised learning (DSL); a thousand human labels rule out Haiku's rising law.

0.38

Usability, User Experience, and Acceptability of the Trail Making Test in Virtual Reality in Younger and Older Adults

cs.HC cs.CY ·2026-10-16 ·Panagiotis Kourtesis, Dante O'Connor, Sarah E. MacPherson
Abstract · 20 pages, 3 figures, 4 tables, supplementary material are included

Immersive virtual reality (VR) adaptations of neuropsychological tests change how a test operates. Before their scores are interpreted, the individuals they are intended to assess should find them usable, acceptable and tolerable, and for older adults this evidence is usually missing. This study evaluated the experience of the Trail Making Test in Virtual Reality (TMT-VR), a head-gaze adaptation of the Trail Making Test that requires no hand-held controller, in 47 younger (20-38 years) and 44 older adults (64-81 years). After the task, participants rated usability, pragmatic and hedonic user experience, acceptability, and cybersickness. The study had three aims: to compare the two age groups on these ratings; to examine whether the ratings were associated with age group, gaming and technology experience, and whether cybersickness was associated with motion sickness susceptibility; and to examine whether the ratings were related to TMT-VR completion and error-adjusted times. Both groups rated the TMT-VR as highly usable, acceptable and pleasant to use. Older adults gave slightly lower usability and pragmatic ratings, but neither the age-group differences nor the associations with age group or previous experience were statistically significant after correction. Cybersickness was low, with 38 of the 91 participants reporting no symptoms; it did not differ between the age groups and was not associated with motion sickness susceptibility. No correlation between the ratings and TMT-VR performance survived correction. The TMT-VR with head-gaze selection is therefore usable, acceptable and well tolerated by older as well as younger adults. Because the ratings were favourable across the whole range of TMT-VR scores, slower or less accurate performance is unlikely to reflect difficulty with the interface.

0.37

From Prefabricated Content to Situated Generation: A Four-Wave Longitudinal Study of AI-Mediated Museum Surfaces and Spaces

cs.HC ·2026-10-16 ·Yibo Meng, Xiujun Chen, Bingyi Liu, Xiaolan Ding
published
Abstract · ACM ISS 2026 PACMHCI full paper; revised manuscript; 5 figures; 3 tables; accepted paper

Generative AI changes a basic condition of interactive surfaces and spaces (ISS): visitor-facing content can be produced during interaction rather than fully authored before it. We report a longitudinal qualitative study with 23 Chinese science museum practitioners, interviewed annually from 2023 to 2026 (retention 23 to 19) and analyzed using reflexive thematic analysis. Participants' accounts moved from anticipated epistemic and operational risks, through backstage experimentation and visitor-facing trials, toward situated governance and renewed questions about the purpose of museum interaction. Following these changes over time shows how generative AI reconfigured the design problems of museum ISS: generated content appeared under institutional authority; inference delay became publicly and bodily experienced; concurrent touch, gesture, and speech created ambiguous multi-user intent; and post-generation editing could leave curators selecting within an AI-defined possibility space. We conceptualize the resulting shift as a movement from prefabricated display to situated generation and formulate four design propositions concerning provenance at the point of display, spatially appropriate fallback, multi-intent negotiation, and curatorial control before generation. These propositions are grounded in practitioner accounts rather than evaluations of implemented solutions.

0.37

Sensory Comfort Across Multisensory XR Prototypes for Autistic Adults

cs.HC ·2026-10-16 ·Yibo Meng, Xiujun Chen, Bingyi Liu, Xiaolan Ding
published
Abstract · ACM ISS 2026 PACMHCI full paper; revised manuscript; 4 figures; accepted paper; Article ISS030

Multisensory XR may support informal learning for autistic adults, but sensory-friendly design often emphasizes reducing input. We conducted a mixed-methods, counterbalanced within-subjects study with 20 Chinese autistic adults who evaluated five complete XR configurations presenting the same popular-science content. Comfort differed substantially across the configurations (Friedman chi-square(4) = 46.39, p < .001, Kendall's W = .58). L3 received the lowest mean rating (2.97/5) and was tied for lowest for 16/20 participants, whereas L5 received the highest mean rating (4.43/5) and was tied for highest for 17/20. Participants described the static configurations in terms of clarity and self-pacing, L3 in terms of abruptness, simultaneous demands, insufficient warning, and limited disengagement, and L4/L5 in terms of gradual transitions, opportunities to disengage, and coordinated narrative cues. The study provides a configuration-level account of sensory comfort and identifies participant-described qualities that can inform future controlled studies of multisensory XR.

0.35

Source Attribution and Audience Uptake in Digital Witnessing: Evidence from a Chinese Diaspora Account on X

cs.SI cs.CY ·2026-10-16 ·Huiqian Lai, Runze Ding
published
Abstract · Accepted for publication in the Proceedings of the 60th Hawaii International Conference on System Sciences (HICSS-60), 2027

Digital-witnessing research explains how citizen testimony becomes publicly visible. Less is known about audience uptake after publication and whether the attributed source corresponds to the form of response. This study analyzes 47,635 posts (April 2022 to May 2026) from Teacher Li Is Not Your Teacher, a diaspora-based account that curates and publishes reports and testimony from inside China for transnational audiences. Among 25,093 witness-relevant posts, 3,601 explicitly attribute their material to ordinary submitters. These posts receive more replies than other witness-relevant posts across adjusted analyses and show a more reply-oriented profile. The primary doubly robust retweet estimate is near zero. Among audited replies to November 2022 user-submitted posts, 91.4% substantively engage the focal material. The pattern is consistent with ordinary-person source attribution presenting testimony as situated and open to public response. The study extends digital witnessing beyond visibility by separating redistribution from the public response that gathers around testimony.

0.35

GCAM-Gym: A Benchmark for Agent-Based Climate Policy Search

cs.CY ·2026-10-16 ·Tianyu Zhang, Zhenghan Tai, Xinyu Wang, Irina Rish, Yoshua Bengio
benchmark
Abstract

Comparing climate-policy search methods requires accounting for differences in policy control, available observations, feedback timing, and simulator budgets. We introduce GCAM-Gym, a Gymnasium-compatible environment that supports full-policy evaluation and stepwise control through a common interface and exposes the components of the policy objective. We compare seven search and control systems in three settings: target attainment and cost, cost-temperature trade-offs, and one-step delayed feedback. Each system is evaluated with ten seeds per setting, for a total of 3,210 attempted GCAM evaluations. Four systems share a budget of 24 attempts per seed; among these, CMA-ES achieves the highest interquartile mean (IQM) of the best returns from completed policy evaluations in all three settings. We release the evaluation archive and analysis tools to reconstruct the reported comparisons without running GCAM.

0.33

Graphlet-based Decentralized Finance Transaction Method Classification

cs.SI cs.CEcs.CYcs.ET+1 ·2026-10-16 ·Célestin Coquidé, Alessio Brini, Natkamon Tovanich
Abstract · 24 pages, 5 figures, 5 tables, 25 references 15th International Conference on Complex Networks and Their Applications

Decentralized Finance (DeFi) transactions involve complex interactions among accounts, smart contracts, and tokens. We investigate whether network topology can identify transaction methods that describe DeFi operations. We extract directed graphlet features to classify the 30 most frequent methods. Our results show that graphlet features complement event log counts in improving classification performance. Using SHapley Additive exPlanations (SHAP), we further derive model explanations that reveal method-specific graphlet structural patterns.

0.33

Revisiting Structure-Based Filtering for AI-Designed Antibodies: Opportunities, Limits, and What Matters

cs.CE ·2026-10-16 ·Langzhang Liang, Juncheng Qian, Xiaojie Ma, Mingxi Zou, Yinghui Xu +2 more
Abstract · 29 pages, 4 figures

AI antibody design can generate far more candidates than can be experimentally tested, making post-generation filtering a central bottleneck. We study whether structure-prediction outputs can help prioritize AI-designed antibodies for experimental validation. We evaluate this problem on 4,184 experimentally characterized antibody-antigen pairs, combining newly generated wet-lab measurements with curated affinity data across diverse antigens and assay settings. We introduce two complementary filtering signals: Regional Interaction Contrast (RIC), which captures support for the predicted antibody-antigen interface, and Non-CDR Antibody Confidence (NAC), which measures antibody confidence outside the CDRs. Across AlphaFold 3, Protenix-v2, and ESMFold2, we find a striking difference in what information matters for filtering. Interface evidence is strongest when comparing candidates designed for different antigens, whereas antibody-level reliability is substantially more informative for ranking candidates within the same antigen. In particular, non-CDR antibody confidence consistently outperforms whole-antibody confidence across predictors, revealing an unexpected source of filtering signal beyond the predicted interface. Finally, we examine the information encoded in the predictor outputs themselves. Predicted structures contain substantial but incomplete binding-related signal, revealing both the opportunities and limits of structure-based filtering: better metrics can exploit current predictors more effectively, but their performance ultimately remains constrained by what those predictors can distinguish.

0.32

The Ethics of Prompt Engineering

cs.HC ·2026-10-16 ·Federico Germani, Giovanni Spitale
positionhot direction
Abstract

Prompt engineering is commonly understood as a technical practice aimed at eliciting useful, accurate, and appropriately structured outputs from large language models (LLMs). Yet prompts do more than optimize model performance: they shape what information is elicited, which assumptions are introduced, and which perspectives are privileged. In this paper, we argue that prompting should therefore be understood as an epistemic and normative act. We distinguish between two conceptions of prompting: knowledge generation, in which the model is implicitly treated as an epistemic authority, and knowledge injection, in which users provide relevant knowledge, evidence, and context that the model can organize, articulate, and extend. We contend that the latter provides a more appropriate model for responsible LLM use, since prompting cannot substitute for subject-matter expertise or the critical evaluation of model outputs. At the same time, prompts inevitably inject values, intentions, and assumptions into the generation process. Building on these premises, we propose five principles for responsible prompting: epistemic humility, explicit engagement with uncertainty and epistemic underdetermination, source sensitivity, avoidance of deceptive framing, and recognition of model limitations and appropriate task allocation. We further argue that responsibility for AI-generated outputs is distributed across users, developers, institutions, and societies, and that some applications cannot be rendered ethical through better prompting alone.

0.31

PresenceBench: An Exploratory Evaluation of Large Language Model Adaptation to User State

cs.HC cs.AIcs.CL ·2026-10-16 ·Peña Arroyo Gallego, Mauro Ruiz Moreno, Alejandro Acien, Sergio Romero-Tapiador, Teresa Arroyo-Gallego
Abstract · 33 pages, 6 figures, 13 tables

Presence-aware intelligence is the capacity of an AI system's intelligence layer to adapt appropriately to the functional state of the person it serves (here readiness, flow, and stability) while preserving task quality and safety. We introduce PresenceBench, an instrument for measuring it. PresenceBench evaluates how the intelligence layer uses a supplied state, not how that state is sensed, and breaks the capability into measurable properties. Two are tested here. Receptivity: does the model take up the state and give a better response than with no signal? Specificity: does any improvement depend on the correct state, rather than on extra structured context? A third, adaptation across states, is a required endpoint not measured in this release. A five-arm causal design answers both questions, against each model's own baseline and an equally structured mismatched-state control, scored by a correctness-gated human paired comparison. The exploratory evaluation covered four models across eighteen single-turn, six fixed three-turn and six simulated natural-conversation scenarios. Supplying a synthetic state, raw or translated into a response policy, gave human-preference point estimates above 0.5 for all four models (Claude Sonnet 5, DeepSeek V4 Pro, Gemini 3.6 Flash, GPT-5.6 Terra), with three of eight 95% intervals excluding chance. Specificity was not resolved: no per-model interval excluded chance, and the design cannot yet separate the state values from a generic instruction to adapt. Policy delivery reduced priced cost and output tokens against each model's baseline; these are resource-use changes, not efficiency gains, because correctness non-inferiority was not established. Safety-floor failures were rare (13 of 360 cases), with a non-significant excess under policy delivery. We report the instrument, the frozen results, and the controls a confirmatory study needs.

0.31

Desirability Confirmation: Relational Sycophancy in AI Intimate Roleplay

cs.CY cs.CLcs.HC ·2026-10-16 ·Huiqian Lai, Runze Ding
published
Abstract · Accepted for publication in the Proceedings of the 60th Hawaii International Conference on System Sciences (HICSS-60), 2027

Research on LLM sycophancy has largely focused on propositional agreement, in which models echo users' beliefs, preferences, or judgments. Adult intimate roleplay presents a different case because what the user seeks may be preservation of a preferred relational position within a generated scene. Using a controlled prompt-based audit of 2,016 outputs from two open-weight LLMs, we identify four recurrent relational moves: legitimating desire, presenting the user as wanted, preserving non-rejection while communicating limits, and keeping intimacy available within fantasy or roleplay frames. Bringing the sycophancy literature into dialogue with sexual script theory, we conceptualize this configuration as desirability confirmation, an output-level pattern that captures how user-pleasing can operate through relational positioning. The findings extend sycophancy research and intimate-AI evaluation beyond binary compliance or refusal by showing how outputs can differ in the ways they communicate boundaries, preserve relational acceptance, and mark intimacy as fictional.

0.30

Over reliant and underprepared: First responder resilience to disruption of operational technology

cs.HC cs.CY ·2026-10-16 ·James Connor, Courtney Crooks, Ryan Shandler
Abstract · 33 pages, 1 figure, 5 tables

The Emergency Services Sector (ESS) is responsible for taking life-saving action in the face of extreme crises. To enhance its response capabilities, the sector has spent the last two decades investing in new technology to improve their operational efficiency. This has created significant interdependencies and a culture among first responders of technical overreliance, raising questions about what would occur if those tools were suddenly disrupted. Several works have discussed the potential for terrorism to intentionally target ESS technology as part of a larger kinetic attack. Despite this, the resilience of first responder agencies to technological disruption and their preparedness for such an attack has largely remained unexplored. To bridge this gap, we implemented a novel scenario-based interview methodology involving senior officials from emergency service agencies across the United States. These interviews explored how each agency prepared for a disruption of operational technology and what factors might influence different levels of preparation between them. Participants were asked to describe their responses to a hypothetical scenario modeled after historical cyber-attacks against ESS infrastructure. The results indicate that while ESS agencies are prepared for short-term technological disruptions, they are likely to struggle with long-term or hybrid incidents. The data further revealed that agencies often lack dedicated expertise in cybersecurity, technology, and emergency management that could otherwise elevate resilience to interruptions of operational technology. These findings emphasize the need for investment in redundant analog systems, increased training for frontline personnel, and a focused effort against loss of specialized technical expertise among ESS agencies.

0.30

On some unresolved issues in aggregate level comparative research evaluation

cs.DL ·2026-10-16 ·Paul Donner
unusual
Abstract

This short conceptual paper recalls to the attention of the scientometrics community some long-standing open methodological problems and proposes new directions that might be taken in renewed engagements with them. These problems are the differential dependency of the global reference score on the included units' observed scores and the dilution of impact measurement accuracy due to collaboration, which causes a global convergence in country citation impact due to international collaborative research. Advantages and disadvantages of existing and new solutions are discussed.

0.27

Designing Human-AI Collaborative Science: The 9A Framework

cs.HC ·2026-10-16 ·Zihan Zheng, Linyue Pan, Xin Lan
Abstract · 19 pages, 2 figures, 2 tables

Artificial intelligence (AI) for Science is often described as a progression from human-directed assistance toward increasingly autonomous AI scientists. Human-AI collaboration offers a further mode in which human and AI contributions substantively shape scientific reasoning, judgments, and the course of inquiry. Existing research addresses work allocation, human-AI coordination, and system evaluation, but how these aspects interact to shape collaborative scientific systems remains insufficiently articulated. We develop the 9A Framework for Human-AI Collaborative Science (HACS) to connect these considerations across three layers. The organization layer specifies what scientific activities are required (Activity), who performs them (Actor), and who holds authority over consequential decisions (Authority). The coordination layer addresses how participants access relevant research information (Awareness), how scientific work advances through iterative proposal, execution, examination, revision, and resolution (Accountability), and when collaborative arrangements should be adjusted (Adaptability). The evaluation layer establishes what constitutes effective system performance (Achievement), how comparisons can be designed to identify the effects of organizational and coordination choices (Attribution), and what evidence supports the resulting conclusions (Assurance). Together, these layers frame HACS as a system-design problem, providing a basis for designing collaborative scientific workflows and for empirically comparing their effectiveness.

0.27

Reinforcement Learning-Based Pricing and Operation for Hybrid Electric and Hydrogen Vehicle Stations

cs.CE ·2026-10-16 ·Lowell Welburn, Yang Chen, Xudong Wang, Abhilasha Saroj, Shaked Regev
Abstract

Electric vehicles (EVs) and hydrogen fuel cell vehi- cles (HFCVs) are promising technologies in the efforts to reduce carbon emissions. Challenges such as limited infrastructure, con- sumer dissatisfaction, and electrical grid strain inhibit progress in the adoption of these technologies. A zero emission vehicle (ZEV) station, capable of handling EVs and HFCVs, addresses these challenges by focusing on maximizing station profit and consumer utility, while minimizing grid impacts. The purpose of this research is to develop a realistic simulation of a ZEV station that is capable of handling real-time pricing decisions enabled by reinforcement learning (RL) and to compare the performance of the RL agent to static pricing methods. After training, the RL agent gains 10.83% greater station profit than the baseline manual parameter search. This work serves as a useful tool for ZEV station owners when handling complex pricing and purchasing decisions.

0.26

Indirect Influences of Artificial Intelligence Advice on Complex Sepsis Treatment Decisions

cs.HC cs.AIq-bio.QM ·2026-10-16 ·Venkatesh Sivaraman, Joel Levin, Eric P. Mason, Victor Talisa, Suchi Saria +4 more
Abstract

Sepsis is a major public health concern, and many sepsis patients die due to ineffective hemodynamic management. Artificial intelligence could support hemodynamic decisions, particularly in complex cases where conflicting information and comorbidities drive uncertainty and practice variation. However, these systems' real-world value depends on not just accuracy, but how they influence the quality of clinician decisions. We first retrospectively evaluated an AI system's recommendations in 6,928 sepsis patients, finding that they improved upon the original providers' decisions overall but had significantly worse performance in complex cases. We then conducted a pre-registered experiment with 228 US critical care physicians to test their use of this imperfect advice in six complex treatment decisions, using one of three advice types designed to vary reasoning autonomy. Participants given simple model recommendations overcame poor advice, choosing optimal treatments in 56.3% of decisions compared to 31.8% if they had always followed the recommendations. They also made significantly better decisions than those receiving a less prescriptive advice type (42.9%). Physicians selectively adapted and improved weak advice, likely because it prompted them to reconsider their assessment. Our findings suggest that improving sepsis care decisions with AI may require accounting for these indirect influences resulting from human-AI collaborative reasoning.

0.26

TRACE: Governing AI-Assisted Research with Principles from Computer Science

cs.CY econ.GNcs.HC ·2026-10-16 ·Amirhossein Zohrehvand, Aldona Kapacinskaite
Abstract · 30 pages, 7 tables

Artificial intelligence (AI) has made it easier than ever to produce research outputs. Adding an agentic tool to a research team's work potentially extends the team's capabilities but calls for an active approach to organizing the research process. The pressing question, therefore, is how research teams should organize their work with this technology. We contend that the question is not new: software engineering and security research have long studied how teams of people working with machines keep control of projects with many contributors. We distill principles from the practices these literatures report and transpose them to AI-assisted research. The principles form TRACE: Track records, Review against sources, Accept independently, Control transitions, and Evaluate corrections. Together, the five principles give researchers a common basis for remaining in control while integrating AI into their workflows. By bounding units of work and attending to how each is shaped, they can help researchers pace AI-assisted work to human attention rather than to the speed at which AI produces content. We also introduce TRACE-OS, an operating system for AI-assisted research projects. It puts the principles into practice through shared instructions, tools, and records that fit researchers' existing workflows. With TRACE, we aim to join a conversation about the governing principles through which the research community can benefit from AI. We contend that a simple enough and useful standard can contribute to harnessing AI capabilities, allowing teams to delegate parts of their work to AI, while also building trust within teams and the broader research community.

0.25

Bad Diagnosis: Analyzing Consent and Data Collection Practices of Mobile Medical Apps

cs.HC cs.CRcs.CY ·2026-10-16 ·Ethan Myers, Lorenzo De Carli, Shantanu Sharma, Ritwik Banerjee, Joel Reardon +1 more
Abstract

Mobile health (mHealth) apps collect personally-identifiable information while handling uniquely sensitive medical data, ranging from mental health records to real-time glucose or heart rate monitoring -- yet they remain poorly understood at scale. We present the first comprehensive analysis of consent and data collection practices across 10,000+ mHealth apps on the Google Play Store, combining static analysis and privacy policy examination. We develop a rigorous operational definition to distinguish health apps from general wellness programs, and design a measurement framework to detect misuse of permissions and violations of consent. Our analyses reveal a troubling ecosystem: 69.15% of apps request at least one permission unrelated to their stated purpose, with over 52.9% apps requesting location data, and 50.5% and 34.9% apps requesting camera and microphone access, respectively -- despite no functional justification based on app descriptions and stated purposes. Most alarmingly, nearly 93% of apps fail to provide users with a consent form, directly violating regulatory requirements such as GDPR; and even among the remaining 7%, 1 in 14 demonstrate a gap between the obtained consent and collection of personally identifiable information. These findings demonstrate an urgent need for stronger enforcement of privacy regulations in mHealth apps. We plan to release our dataset and measurement tools for continued monitoring of this critical space.

0.25

CROPS: Cosserat-Rod-Based Co-Design of Soft Slender Programmable Material Systems for Biomedical Applications

cs.CE cs.RO ·2026-10-16 ·Aaditya Chandrasekhar, Dex Doksoo Lee, Hyunwoo Kwon, Shicheng Li, Wei Chen
Abstract · Submitted to Finite Elements in Analysis and Design

Slender structures made of stimuli-responsive materials represent an emerging class of programmable material systems, with applications in soft robotics, minimally invasive surgery, wearable electronics, and deployable structures. In biomedical applications, the on-demand design of slender structures, encompassing both their material layout and actuation, is essential to harnessing their large deformation and multiphysics actuation to meet patient-specific design requirements. However, owing to geometric and material nonlinearities, their structure-actuation-response relationship is highly nonintuitive and computationally demanding. This often makes trial-and-error and conventional design approaches challenging for addressing patient-specific needs. To this end, we present a Cosserat-rod-based design optimization framework for the autonomous design of programmable slender structures. Leveraging the Cosserat rod representation, the end-to-end differentiable framework is demonstrated on two case studies: (i) design of swellable stents that morph to the shape of carotid vessels; (ii) co-design of the remnant magnetization and the applied field of magnetically-active catheters to reduce the force exerted on the patient-specific vessel wall during insertion. Both self-contact and external contact with surrounding structures are accounted for. We consider two actuation mechanism, namely the swelling of gel-elastomer composites and the magnetic actuation of hard-magnetic composites. The proposed framework presents an autonomous design platform of soft slender material systems that support large deformations and multiphysics actuation toward personalized biomedical design.

0.25

Can AI interview? Comparing AI-led to Human-led Chat Interviews

cs.CY cs.CLcs.HC ·2026-10-16 ·Semra Guven, Tobias Gårdhus, Andreas Bjerre-Nielsen, Anna Rogers, Hjalmar Bang Carlsen
code/data
Abstract · 37 pages, 3 figures, 6 tables. Submitted to Socius. Code and data: https://github.com/ainterviewer/can-ai-interview

Large language model chatbots have been proposed as AI-interviewers for collecting qualitative interview data at scale, potentially bridging social science's longstanding divide between the breadth of surveys and the depth of qualitative interviews. This paper provides the first direct empirical comparison of data quality in AI-led and human-led interviews. In a randomized experiment (N = 41), university students were interviewed via synchronous text chat by either trained non-expert interviewers or a locally hosted AI system; we evaluate exposure, specificity, and relevance. Median interview duration was similar across conditions, but AI-interviewers asked more than twice as many questions per interview and elicited roughly 65 percent more respondent text per interview. Per-question responses were longer with human interviewers, though not after multiple-testing correction; only AI-condition participants skipped questions. Responses were almost uniformly on-topic in both conditions; for specificity we detect no difference, but the design cannot exclude one of up to three quarters of a point. These results were obtained with a small, locally hosted open-weight model (7B parameters), which addresses concerns about data security, GDPR compliance, and scientific reproducibility. Our findings support the viability of AI-interviewing for large-N qualitative data collection.

0.25

MultiPASE: Shape-Aware Comparative 3D Genome Analysis

cs.HC q-bio.QM ·2026-10-16 ·N. Nipu, G. Johnson, C. Roth, D. Rogers, S. R. Starkenburg +2 more
Abstract

3D genome structures reconstructed from Hi-C data provide valuable insights into chromatin organization, but comparing them across conditions and time points remains challenging because of noise, variability, and arbitrary orientation. State-of-the-art comparison approaches often rely on contact-matrix-based summaries, which do not preserve the spatial geometry of the reconstructed structures and make integration with gene expression and chromatin accessibility difficult. There is therefore a need for robust structural alignment and spatial abstraction methods that support integrated analysis across conditions and time points. We present MultiPASE, a computational pipeline for comparative analysis of reconstructed 3D genome structures represented as gene-level spatial point clouds. MultiPASE aligns point clouds into a shared coordinate frame using coarse-to-fine rigid registration with PCA-based axis disambiguation and robust ICP refinement, then summarizes spatial organization using two complementary shape abstraction methods that produce stable, nested, multi-scale representations from dense core regions to broader structural extent. Using Hi-C-derived reconstructions from viral infection experiments, we show how the approach reveals condition-dependent structural changes that are difficult to interpret from raw 3D structures alone. By integrating spatial structure with gene expression and chromatin accessibility, MultiPASE supports interactive analysis of spatial organization and gene relocation patterns. A quantitative validation, reported in the Supplementary Materials, shows that the proposed shape abstractions preserve non-convex features, remain stable across samples, and that the alignment pipeline is necessary for meaningful comparison. A usage case and a usability evaluation with eight participants demonstrate the pipeline's effectiveness for comparative structural analysis.

0.24

Trinketry: Materializing Elements to Anchor Traceable Design Histories

cs.HC ·2026-10-16 ·Wen-Fan Wang, Yi-Ting Chiu, You-Yi Hsieh, Bing-Yu Chen, Max Kreminski
Abstract

Generative tools enable visual designers to rapidly explore diverse design directions---but as flawed generated outputs proliferate, designers may lose track of promising ideas and the context of their development. We introduce trinkets: user-preserved design elements that function as reusable materials, retrieval queries, and navigational anchors for tracing how ideas emerge, transform, and recur throughout design history. We implement this concept in Trinketry, a visual exploration canvas. In a within-subjects study with 12 concept designers, Trinketry supported focused element-level exploration and greater perceived control than a simple branching image generation history baseline. A follow-up study with six returning designers examined delayed design history sensemaking, continuation of work, and reconstruction of others' processes. Compared with whole-image histories, Trinketry supported more precise reconstruction of process structure, turning points, and goals, while helping designers recover overlooked materials and continue exploration. However, opportunities remain to augment trace representations with explicit statements of design rationale.

0.22

Modeling Group Socialization in Online Communities

cs.SI cs.HC ·2026-10-16 ·Vinay Koshy, Iris Zhong, Karrie Karahalios, Benjamin Mako Hill, Amy X. Zhang
Abstract

Newcomers to online communities must learn a group's norms to participate successfully. This process is known as socialization, and has been the subject of considerable study in the online communities literature. Foundational work on group socialization in social psychology, however, suggests that socialization is actually a bidirectional process: individuals and communities adapt to each other's needs. In this work we develop a formal quantitative model of socialization, and present the first large scale empirical validation of the bidirectional socialization hypothesis using Reddit data. More specifically, we focus on linguistic socialization, and model users' comments in online communities as a mixture of in-group language and language from ``background'' communities they previously participated in. On a dataset of 112 subreddits (7.8 million tokens) find that our bidirectional model makes predictions that better generalize across users and time than two unidirectional socialization models. Furthermore, we conduct a qualitative analysis in which we demonstrate that our model is able to extract interpretable socialization patterns from the data that align with the existing literature. Specifically our model is able to learn that users conform more to group norms in topically niche communities with stricter content moderation rules, and that long time users conform more than newcomers. We emphasize that the claims made in this paper are explicitly falsifiable -- we are staking out a clear claim about what the socialization process looks like, and encourage future work to challenge our model.

0.19

Evaluating Predictor-Corrector Dependence in Hard-Constrained Climate Emulators

cs.CY cs.CE ·2026-10-16 ·Tianyu Zhang, Xinyu Wang, Zhenghan Tai, Xiao-Wen Chang, Yoshua Bengio
Abstract

Climate emulators often enforce physical constraints by correcting each state predicted by a neural network, but evaluating only the corrected forecast conflates predictor skill with the effect of the correction. We introduce an evaluation framework that separates these contributions by comparing corrected forecasts, pre-correction predictions along corrected trajectories, and autonomous rollouts with correction disabled. We further cross correction during training and inference to distinguish its immediate effect from how it shapes the learned predictor, and use matched nonphysical controls to test whether improvements are specific to the physical correction. In ACE2, correction substantially improves physical consistency. In a three-seed ACE2 pilot, models trained with correction produce less accurate uncorrected precipitation forecasts at short horizons than models trained without correction, but more accurate forecasts at 10 and 30 days. This contradicts our preregistered hypothesis that the deficit would grow with forecast horizon. In a separate nine-seed experiment, directly supervising the uncorrected output reverses its precipitation forecast deficit at 72 hours while maintaining the accuracy of corrected forecasts at 24 hours. In ConCerNet's heat-equation system, projection-trained models have higher uncorrected trajectory-based vector-field RMSE than models trained without projection, while restoring the physical projection recovers accuracy far more effectively than a matched nonphysical control. Together, these results show that hard corrections can reshape the predictors trained alongside them, creating dependencies that are not visible from corrected forecasts alone. Hard-constrained emulators should therefore be evaluated by separately measuring predictor accuracy, correction effects, and their interaction during training.

0.18

A Large Language Model LLM Approach to COVID-19 Fake News Detection

cs.CY ·2026-10-16 ·Muhammad Arif, Samah Fodeh-Jarad, Elyas Irankhah, Uroosa Fatima, Lama Ayash +1 more
Abstract · Pages 15, 10-15 Figures

Text classification is a widely studied problem in information retrieval and data mining, with applications ranging from healthcare diagnosis to the detection of COVID-19-related fake news. Social media platforms generate vast volumes of user-generated text, and the rapid spread of misinformation during the pandemic has made automated fake news detection an important task. In this study, we apply pretrained transformer-based transfer learning (TL) models and large language models (LLMs), namely BERT-base, ALBERT-base, XLM-RoBERTa-base, DistilBERT, GPT-2, XLNet, and T5, to the COVID-19 English Tweets dataset, which contains 6,420 tweets labeled as real or fake. To further improve performance, we apply six late-fusion optimization techniques: weighted fusion, particle swarm optimization (PSO) fusion, the Nelder-Mead algorithm, Powell's method, the Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm, and trust-region constrained optimization. Models are evaluated using accuracy, precision, recall, and F1-score, and confusion-matrix heatmaps are used to visualize the predictions of each model. GPT-2 achieved the best performance, with an accuracy of 0.9631, precision of 0.9612, recall of 0.9610, and F1-score of 0.9600, while Nelder-Mead fusion achieved the best result among the fusion methods with an accuracy of 0.9610. These results highlight the value of LLMs for COVID-19 fake news detection. We conclude by outlining future directions, including multiclass classification, multilingual datasets, and real-time social media streams.

0.18

Validating LLM-generated political networks with survey-reported partisan exposure

cs.SI cs.AIphysics.soc-ph ·2026-10-16 ·Marco Tonin, Michele Tizzoni, Bruno Lepri, Luca Maria Aiello
benchmark
Abstract

Large language models (LLMs) are increasingly used to simulate human behavior in social science research, including decisions, interactions, and social relationships. However, in network formation tasks, existing evaluations often focus on aggregate network properties, while networks with plausible structural properties may still fail to reproduce individuals' social and political environments. Here, we generate LLM-based political networks using personas sampled from the nationally representative ANES 2020--2022 Social Media Study and use respondents' survey-reported partisan exposure as an empirical benchmark for the political environments generated for the corresponding personas. We first characterize the structural properties and political assortativity of the generated networks, and then compare ego-level partisan exposure with the survey. We find that the structure and political assortativity of the generated networks vary across experimental configurations. More importantly, when comparing generated partisan exposure with the survey, some models generate stronger separation between Democratic and Republican egos than reported in the survey, with higher same-party and lower cross-party exposure, while others mainly underestimate Republican exposure across ego groups. Moreover, networks with similar political assortativity can still generate different partisan environments around the same personas, including among networks with plausible structural properties. Finally, generated networks show group-specific errors, under- or overestimating exposure to either Democrats or Republicans for specific social groups. These findings show that evaluating LLM-based social simulations only through aggregate outcomes can hide errors in the local environments through which those outcomes emerge, highlighting the importance of validation at both aggregate and individual levels.

0.17

Regulating for an AI Commons: Adopting, sustaining, and defending people-first innovation trajectories

cs.CY ·2026-10-16 ·Abhineet Nayyar, Mallory Knodel, Kimberlee Weatherall, Gunn Jiravuttipong
position
Abstract

Technological principles of openness, interoperability, and decentralization are integral to building an AI economy that is accountable to the public and truly responsive to local needs and constraints. However, Big Tech's firm control over essential elements of the stack (such as chips, cloud, models, data, and talent) limits valuable alternatives from reaching the scale necessary for sustained growth. Worse still, hyperscalers continue to enclose and extract value from publicly funded and volunteer-driven technologies without adequately distributing these returns. We see consolidated power within a few entities and "enshittification" of once-transformative innovations. Embedding ethical micro-level tools (like open data repositories or small-language models) into workflows is promising, but without a consistent innovation strategy, they also often end up perpetuating and being appropriated by the same exploitative trajectories. Instead, this whitepaper outlines a three-pillared holistic and intentional policy approach that i) actively pushes for the adoption of cooperative technologies in existing and emerging systems, ii) sets up collective institutional structures to sustain and maintain them in the long-term, and iii) introduces regulatory proposals that limit private entities from disempowering them through uncontrolled enclosure and extraction. Through desk research and purpose-specific case studies, we argue for the importance of such an approach in the Global South, where structural constraints around infrastructure, financing, and policy governance make it difficult for AI commons and cooperative AI tools to emerge and survive. The paper concludes with key open questions for campaigners, policy advocates, researchers, and grassroots organizers to consider and explore while working towards peoples' solidarity within and between Global South countries.

0.15

A Multimodal Framework for Measuring Creator Communication on Short-Form Video

cs.SI cs.CLcs.CY ·2026-10-16 ·Julie R. Ancis, Michael Fire, Omer Tarshish
Abstract · 17 pages, 3 figures, 7 tables

Short-form video combines speech, imagery, and on-screen text, creating challenges for systematic content analysis. We present a five-stage multimodal framework for measuring creators' themes and communication strategies at scale. The framework integrates platform metadata, speech transcripts, and vision-language descriptions of sampled frames, then applies large language models to generate predefined indicators and open-ended labels. An audited keyword scheme consolidates thematic labels, while creator-weighted aggregation gives each account equal weight regardless of posting volume. We demonstrate the framework on 9,852 TikTok videos from sixteen creators, retaining 7,311 videos for analysis across seven themes and seven communication indicators. Automated profiles reveal substantial differences in creators' use of education, humor, and responses to current events. Validation against two independent expert annotators yielded a mean framework-expert agreement of Cohen's $κ= 0.40$ across the six binary indicators, close to the $κ= 0.42$ the two experts achieved with each other, though agreement varied substantially across individual indicators. The framework provides a modular approach to comparative analysis of creator communication, combining scalable annotation with transparent aggregation and construct-specific validation.

0.15

AlphaLedger: Portfolio-Grounded On-Policy Distillation for Regime-Aware Alpha Mining

cs.CE ·2026-10-16 ·Xiang Li, Zikai Wei, Yiyan Qi, Wanyun Zhou, Xiang Liu +5 more
Abstract

Hosted LLMs enable automated alpha mining but incur recurring API costs and expose research context to external providers. Training a local generator requires supervision aligned with portfolio utility: a predictive factor may add little after redundancy and trading costs. This incremental value is assessed only after the factor program is complete, whereas generation proceeds sequentially. Deployment must also adapt to evolving stock-specific evidence and changes in factor effectiveness across market regimes. We propose AlphaLedger to address these challenges. Economic-context on-policy distillation (EC-OPD) conditions a frozen teacher on completed factors and their cost-adjusted marginal portfolio contributions, then distills its legal-action distributions along the student's own trajectories. The deployed generator requires neither teacher calls nor outcome inputs. EviPool constructs a dynamic stock universe from lagged multi-source evidence, while a regime-residual gate (RRG) reweights factors using matured, state-matched returns with shrinkage and bounded deviations from equal weighting. On Chinese A-shares, AlphaLedger achieves a 55.4\% annualized net return, a Sharpe ratio of 2.06, and a maximum drawdown of 10.8\%. In a common CSI500 universe, its annualized net return exceeds the strongest baseline by 43.6\% (31.6\% vs.\ 22.0\%), with a 13.5\% higher Sharpe ratio.

0.13

"To Boldly Go Where No One Has Gone Before" What Has AI Let Developers Unlock?

cs.HC ·2026-10-16 ·Rudrajit Choudhuri, Jenna L. Butler, Christian Bird, Carmen Badea, Brian Houck +1 more
Abstract

Measuring AI by how much faster developers work is like judging electricity by how much faster it spun a steam-era factory. A fuller measure asks what new work becomes possible. We surveyed 331 software developers at Microsoft about what AI lets them do that they could not do before and analyzed their answers using reflexive thematic analysis with a human-in-the-loop, multi-model AI council. Developers described doing familiar work faster (Faster, 45%), undertaking work previously beyond their reach (Further, 36.9%), and taking on new kinds of work centered on building for and directing AI agents (Afresh, 12.4%). Across these accounts, learning and scaffolding ran in both directions: developers learned new skills with AI while externalizing their tacit knowledge as the context and standards that guide AI's work. Drawing on Vygotsky's zone of proximal development, we frame developers and AI as joint learners whose zones meet at the skill-judgment nexus: AI can supply skill faster than it can reliably judge its own work, and developers must continuously bridge that gap. Whether they do sets them on one of two paths. On one, developers' judgment grows alongside AI's skill, and they codify what they learn so the loop keeps turning. On the other, they become hollow orchestrators, approving work they can no longer judge and losing the identity and meaning they once drew from it. Based on our findings, we offer a two-way scaffolding playbook to help developers, teams, and tool builders take the first path.

0.12

Water-Energy-Data nexus reveals sustainability gaps and priorities for AI infrastructure

cs.CY ·2026-10-16 ·Jincheng Huang, Guoan Xiao, Xin Lan, Caichun Yin, Zhimeng Jiang +2 more
Abstract

The rapid growth of artificial intelligence (AI) depends on interconnected water, energy, and data systems spanning infrastructure life cycles and diverse geographies, with the potential to intensify water and energy insecurity, exacerbate social inequities and lead to cascading sustainability issues globally. Yet assessments commonly examine them separately, limiting our understanding of where resource pressures arise, how they are transferred across space, and who bears their consequences. To address these challenges, we established a water-energy-data (WED) nexus framework that captures the spatial and life-cycle dependencies of AI infrastructure, and use large language models to synthesize 12,825 water- or energy-related studies across four resource-intensive AI infrastructure activities: mineral extraction, semiconductor manufacturing, data center operation and electronic-waste processing. The evidence map identifies a mismatch between the interconnected activities supporting AI infrastructure and boundaries used to assess them. Researchers assess operational energy more than water use and other indirect demands, while interactions across locations and life-cycle activities remain understudied. These gaps obscure how resource demands shift among infrastructure nodes and regions, and how data-enabled benefits compare with their underlying costs. The WED nexus provides a foundation for integrated assessment of spatially metacoupled systems, while the evidence map can inform priorities for managing the sustainability implications of the AI boom.

0.10

Wait for It: A Component-Level Analysis of Latency Feedback Mechanisms for LLM Agents in VR

cs.HC ·2026-10-16 ·Pratham Arora, Abhinav Lodha, Siddharth Siddharth
Abstract

Delays in AI responses pose a significant challenge to immersive Virtual Reality experiences. While previous research demonstrates that feedback can reduce the perception of waiting, we lack understanding of which types work best and why. We conducted a within-subjects study with 18 participants to systematically compare three isolated feedback approaches during VR agent interactions: verbal fillers (e.g., um..., hmm...), visual indicators (loading spinners), and embodied gestures (thinking animations). Our results revealed that embodied gestures significantly outperformed visual loading indicators across multiple dimensions, making agents feel more lifelike and socially present. Interestingly, explicit visual feedback like loading spinners actually degraded perceived humanness and broke immersion by forcing users to shift from social to technical mental models. These findings suggest that natural, behavior-based feedback maintains the illusion of conversing with a thinking entity, while technical indicators expose the computational substrate. We present empirically-grounded design guidelines for creating more engaging AI-powered VR agents.

0.08