CS Paper Radar

Daily listing

2026-10-22

764 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 252

PaperFormalizer: Author-in-the-Loop Autoformalization of Research Papers in Lean

cs.AI cs.LO ·2026-10-22 ·Jiyuan Tan, Jose Blanchet, Vasilis Syrgkanis
code/datahot direction
Abstract

Even a carefully proofread paper can leave small gaps in its proofs or its stated assumptions. We introduce PaperFormalizer, a Lean-based agentic pipeline that helps authors find and close such gaps in their own papers and turn each paper into a Lean-verified version. Starting from the LaTeX source, PaperFormalizer transcribes the paper into a dependency graph of statements, audits each statement in natural language for faithfulness, ambiguity and correctness, and then formalizes and proves the statements in Lean against Mathlib, optional domain libraries and the author's personal toolkit. Only the author knows what a statement is meant to say and how a formalization gap in a result is best fixed. The pipeline therefore stops and asks the author whenever a statement admits more than one reading, presents every repair, in natural language and TeX symbols, as a proposal that the author can accept, decline or redirect, and leaves to the author every decision that could potentially change what the paper claims. The output is an annotated and revised version of the paper, a report of the changes, and Lean code in which each declaration is linked to the statement it formalizes. The author's toolkit grows as more papers are formalized, and authors can share their verified papers so that researchers who build on them can reuse their Lean code. By making verified papers practical to produce, PaperFormalizer aims to make research that rests on mathematical proofs more robust. PaperFormalizer runs inside a coding agent (Claude Code or Codex) and is available at https://github.com/Jiyuan-Tan/PaperFormalizer. Papers with Lean code shared by users are available at https://paperformalizer.org.

0.71

PaperScorer: How Predictable Is Paper Acceptance?

cs.LG cs.CV ·2026-10-22 ·Sachin Konan, Jonathan Liu, Zhuang Liu
top venuebenchmark
Abstract · Accepted to NeurIPS 2026

Automated research is largely evaluated by benchmark performance, but the value of research rarely reduces to a single score. Humans judge a work's holistic contribution through peer review, which is hard to model because it is noisy and depends on many factors. However, what if, given only the paper, we could predict its peer-review decision directly? We introduce PaperScorer, a family of models that predict a paper's accept or reject decision from its extracted text or page screenshots, training only on the decision after removing author identity and formatting shortcuts. Vision models are the most predictive (68% accuracy on ICLR papers and 77% on ML papers from arXiv), beating text-only and frontier reasoning models. Unlike plain text, page images retain figures, layout, and presentation. This signal is robust, surviving smaller models and even compressing the whole paper into a single image. But accurate predictions are not evenly spread. PaperScorer is strongest on the lowest and highest rated papers, while papers rated in the middle stay unpredictable for our models and frontier systems alike. Additionally, its confidence is calibrated, so its most confident predictions are also its most accurate. Overall, PaperScorer shows how much of the decision a paper carries on its own, marking a first step toward AI-driven research evaluation.

0.64

Solvable models of learning to pursue a moving target

cs.LG ·2026-10-22 ·John J. Vastola, Kanaka Rajan
publishedbenchmark
Abstract · Accepted to RLC 2026, to be published in the Reinforcement Learning Journal

Agents tasked with intercepting a moving target must learn how to reach it, ideally as efficiently as possible. As observers of animal pursuit behavior have noticed, different strategies are possible: on one extreme, the agent reactively moves toward the target's current location; on the other, the agent predicts the target's future location and moves directly there. Motivated by the desire to understand how such strategies might be learned, we introduce a continuous-time linear-quadratic model of open-loop target pursuit whose optimal control strategies interpolate between these possibilities. Usefully, a small number of interpretable parameters control which type of strategy is optimal, and it is possible to derive closed-form solutions for optimal strategies, policy learning dynamics, and value learning dynamics. Exploiting our model's linear structure, we find that the time scales of learning precisely correspond to the eigenvalues of certain matrices, and that relevant eigenvalue spectra generically have a gap. We show that this gap indicates that agents tend to learn how to reach their target before optimizing their movement. Our results provide a detailed mathematical characterization of pursuit behavior and its learning dynamics, which can both serve as a benchmark for empirical work, and as a foundation for more elaborate theoretical treatments of pursuit.

0.61

The Value of System 1 Models in Medicine

cs.LG ·2026-10-22 ·Xiufeng Yang, Konstantinos Vrettos, Zeineb Ben Chaaben, Fabian Drexel, Jannik Lübberstedt +15 more
Abstract · 27 pages, 6 figures; includes Supplementary Appendix (3 figures, 5 tables)

Many clinical decisions are a choice among a few fixed options, taken quickly and in large numbers. Such decisions also need an uncertainty estimate to know when to defer to a clinician. Generative language models answer slowly in free text and state their confidence poorly. In contrast, System 1 models answer zero-shot with a probability for each answer. However, their performance across clinical tasks is unknown. Four System 1 models, two proprietary and two with open weights, were evaluated zero-shot on 86 public tasks from 14 clinical domains in 20 languages, spanning examination questions and medical imaging. Tasks were assigned to six capability tiers before analysis. The primary outcome was the area under the receiver-operating-characteristic curve (AUROC). Secondary outcomes included calibration, language effects and comparisons with published results and generative models. Across 179,111 cases, the mean AUROC of the best model per task was 0.94 for medical knowledge, 0.93 for text understanding, 0.83 for clinical judgement from text, 0.89 for image perception, 0.84 for 2D image findings and 0.69 for volumetric findings. Imaging tasks scored 0.09 (95% CI, 0.06 to 0.11) lower than text tasks (P < 0.001). The best model matched or exceeded the best published result on 13 of 26 text tasks and on 8 of 35 imaging tasks. Triage nurses outperformed the best model (accuracy, 0.85 vs. 0.56). One decision took 0.34 to 0.44 seconds and 24 to 81 joules on open-weight models run locally. Accuracy on identical questions fell by up to 39 percentage points from English to Yoruba. Generative models of the same family were 11 to 76 times slower, with 3 to 5 percentage points higher balanced accuracy. System 1 models decide quickly, cheaply and with usable confidence when the answer is stated in a text or recalled from medical knowledge, but not when it must be judged or read from an image.

0.53

Scaling Laws for Generative Reward Models

cs.AI ·2026-10-22 ·Angel Raychev, Nathan Lile, Rafael Rafailov
code/data
Abstract · 28 pages. Code and research artifacts: https://github.com/RaychevAngel/generative-reward-models

We study generative reward models (GenRMs) as online preference providers for policy optimization in creative writing. Across Qwen3 reward-model sizes from 0.6B to 14B, Thinking judges trained with GRPO have higher static Gold-label accuracy than Answer-Only judges trained with SFT. This ranking often reverses when the judges train policies through online DPO: Answer-Only rewarders yield higher peak and top-five-average policy Elo in seven of eight matched settings under a fixed, learned 32B Gold evaluator. Intermediate Thinking-judge checkpoints can also outperform later, more statically accurate checkpoints as rewarders. Sweeps over rewarder size, policy size, training duration and DPO regularization show substantial benefits from reward-model capacity, including judges larger than the policy. We analyze trajectories with evaluator-specific Elo arenas, conditional match-resampling intervals and descriptive scaling fits. These results demonstrate an evaluator-rewarder gap in the studied pipelines: fixed-distribution judging accuracy does not determine downstream optimization quality. The comparison changes judge-training objective as well as response format, and the main ranking is conditional on its learned reference rather than a universal human-quality result. A limited follow-up using a different learned judge is inconclusive overall.

0.52

On the Token Collapse Phenomenon in Textual Graph Soft Prompting

cs.LG ·2026-10-22 ·Andrea Giuseppe Di Francesco, Rishabh Jain, Franco Maria Nardini, Nicola Tonellotto, Pietro Liò +1 more
code/data
Abstract · Preprint under review

Recent methods for injecting graph structure into LLM prompts include graph soft prompting, in which a graph neural network (GNN) learns graph-specific tokens that capture relational information among entities in textual prompts. G-Retriever combines graph soft prompts with a verbalized description of the same graph to strengthen the textual prompt signal. We conduct an in-depth analysis of G-Retriever and uncover a pathological behavior we term token collapse. Across three benchmarks, three graph encoders, two LLM families, and both frozen and LoRA-adapted models, G-Retriever's graph tokens become almost indistinguishable across inputs. We provide evidence that the same failure is inherited by subsequent methods, even with multi-token pooling. Under the standard setup, swapping a token with the dataset mean, a token from another example, or one derived from a corrupted graph produces no resolved answer-level effect in accuracy. Moreover, a learned constant prompt that never accesses the graph matches or surpasses G-Retriever on all three benchmarks. These findings indicate that, rather than conveying graph content as intended, the collapsed token functions primarily as a generic soft prompt. We also examine conditions linked to this collapse and propose a two-stage procedure that yields non-collapsed tokens while maintaining competitive performance in most cases. However, token variability alone does not prove that the model uses graph information: via interventions and decoding probes, we show that depending on the configuration, a token can reflect the question, an answer prior, or graph-derived content that the LLM may or may not use. Overall, our results offer a practical approach to evaluating graph-token alignment rather than inferring it from task accuracy alone. Code to reproduce our experiments is available at https://github.com/difra100/Token_Collapse_KGQA.

0.51

Deep Learning for Random Objects in Metric Spaces: Beyond Manifold Assumptions

cs.LG stat.MLmath.STstat.ME ·2026-10-22 ·Marcos Matabuena, Xiaokai Luo, Oscar Hernan Madrid Padilla
hot direction
Abstract

Deep learning (DL) methods have substantially improved predictive performance across a wide range of downstream prediction tasks, enabling new discoveries in multiple domains. However, principled DL methods for non-Euclidean random objects remain relatively scarce, despite the growing importance of modern biomedical outcomes taking the form of probability distributions, random graphs, functions, and other random objects in metric spaces. Motivated by a new generation of digital and clinical biomarkers, we propose a deep learning framework for regression with random objects in metric spaces. Our approach operates directly on the metric structure of the response space, avoiding restrictive manifold or Euclidean embedding assumptions imposed by many existing methods. At the same time, it retains modern statistical guarantees under the stated statistical assumptions. We establish statistical consistency and convergence rates for the proposed estimator, and characterize conditions under which it can mitigate the curse of dimensionality for hierarchically compositional regression functions. Extensive simulation studies demonstrate competitive predictive performance, with gains over several competing approaches in multiple Euclidean and non-Euclidean configurations. Finally, we demonstrate the practical utility of the proposed framework in two digital health case studies with distribution-valued outcomes in Wasserstein spaces, using high-resolution accelerometry and continuous glucose monitoring data. Embedding-based deep-learning methods may be less suitable when no informative low-dimensional representation exists or when the learned embedding does not adequately preserve the intrinsic geometry of the response space.

0.51

VibeGame: Prompt-to-Game Development with AI-Native Engine and Self-Evolving Adversarial Agent Team

cs.MA ·2026-10-22 ·Wenbo Hu, Ken Li, Jiazhe Wei, Yukang Cao, Weiyi Hong +11 more
community pick
Abstract

Large language model agents have enabled software development through natural-language interaction, but extending this paradigm to end-to-end game development remains challenging. Building a playable game requires coordinating code, scenes, assets, animations, physics, and configurations, while allowing developers to validate their integrated behavior through interactive testing. We introduce VibeGame, a framework for creating complete 2D game projects from natural-language prompts and optional reference images. Its AI-native game engine represents project state as schema-validated structured text and provides frame-synchronous runtime control, enabling agents to inspect game states, execute actions, and verify behavior at their own pace. Built upon this interface, we design an Adversarial Agent Team of eight specialized agents to separate implementation from review, so that problems are found and fixed through multi-round loops. For each conducted and accepted project, we further introduce a training-free self-evolution mechanism to distill it into reusable skeletons, modules, and collaboration contracts for future development. We evaluate prompt-assets-to-game (PA2G), where agents are given an asset library, and prompt-to-game (P2G), where agents generate their own art. On a GameCraft-Bench subset, VibeGame in P2G consistently outperforms a range of agent harnesses in either P2G or PA2G with the same underlying model, including GPT-6.1-Sol.

0.50

The Value of Memory in Language Agents

cs.AI cs.CL ·2026-10-22 ·Zehao Lin, Erenkai Zhu, Helan Hu, Xixuan Hao, Chunyu Li +4 more
Abstract · 35 pages, 4 figures, 16 tables

Language agents carry information from past experience into future tasks whose demands may be unknown or change over time. We develop a decision-theoretic account of the capability available in history, retained in storage and realized through use, together with the costs of recovery and revision. Drawing on comparison of statistical experiments, information acquisition and the value of computation, we express these values in units of avoided task loss. Recovery and revision define attainable risk-cost profiles and preservation across uses. A cost-aware simulation condition and five minimal counterexamples establish how information timing, access, computation, feedback and demand shape these comparisons; a pointer with zero answer mutual information can reduce the cost of exact recovery. Allocation and recovery criteria identify when gains pay for estimation, selection and evidence access. Complete-split studies on LoCoMo, LongMemEval-S and PersonaMem cover 3,075 questions and two readers. They show complementary value from saved state and recovered evidence, task-dependent performance orderings, and the role of reuse in amortizing construction costs. Multi-project workflows connect correction to preservation and validation costs. Together, the theory and studies support decisions about what to retain, which recovery options to preserve, and how to adapt memory as evidence and objectives evolve.

0.50

Does LeWorldModel Preserve Local Action Geometry?

cs.LG cs.RO ·2026-10-22 ·Vasil Vasilev, Siddharth Nayak
publishedcommunity pick
Abstract · 13 pages, 3 figures. Accepted as a poster at the NeurIPS 2026 Workshop on Symmetry and Geometry in Neural Representations (NeurReps), Extend…

We examine how closely pretrained LeWorldModel (LeWM) reproduces the effects of small action changes. Using matched simulator and model responses, we measure action sensitivity with Gram matrices at three perturbation scales. Across nearly a thousand states from Reacher and OGBench Cube, the most sensitive action directions differ by median angles of about $36^\circ$ and $42^\circ$, respectively, at the smallest scale. Increasing the perturbation tenfold reduces median normalized Gram discrepancy by roughly 70\% on Reacher and 47\% on Cube. Geometry measured in pixels and embeddings varies more across scales than the chosen physical-state geometry. On held-out directions, environment Grams achieve median embedding response errors of 17\% and 28\%, compared with 26\% and 38\% for an isotropic reference with the same trace. LeWM's errors are higher at this scale. These findings support scale-specific evaluation of action sensitivity; its effect on control remains an open question.

0.48

TwinStore: A Synthetic Consumer Platform Using Vision--Language Models

cs.AI ·2026-10-22 ·Minhyeok Lee
Abstract

We present TwinStore, a source-available synthetic consumer platform for physical retail built on vision--language models. TwinStore turns a store into a digital twin---from 82 ready-made formats, a floor-plan image or phone photos of its shelves---and fills it with synthetic shoppers who plan a trip, walk the aisles and decide what to buy by looking at rendered shelves with packs, price tags and promotion signs. Every trip ends in a receipt, so a layout, price, pack or promotion can be tried on the same shoppers before it reaches the store. Around this core, TwinStore provides 3D playback, paired comparisons and layout optimization, shopper interviews, a trade-area model, counterfactual edits of real shelf photos, a marketing studio with eleven tools, and an assistant that turns a plain sentence into a drafted study or an answer grounded in a run's own numbers. In a held-out validation on eleven public datasets, simulated receipts approach the households' own natural variation (category-mix divergence 0.017--0.018 against a real drift of 0.012), price elasticities fall inside the range measured in the field, and three independent decision models agree. TwinStore runs on cloud or on-device models and is distributed with installers for Windows, macOS and Linux.

0.47

Cost-Aware Abstention for LLM-Based IT Ticket Classification

cs.LG cs.AIcs.CL ·2026-10-22 ·Sachin Gupta
publishedhot direction
Abstract · Accepted at the 38th IEEE International Conference on Tools with Artificial Intelligence (ICTAI 2026). 9 pages, 5 figures, 9 tables

Wrong automatic routing of an IT support ticket can cost more than human triage. We evaluate cost-aware abstention as an offline, cost-conditional readiness assessment: select a routing policy on calibration data under explicit costs, then evaluate that frozen policy once on held-out test data. On a public 47,837-ticket, eight-class corpus, we compare four trained non-LLM classifiers with four large language models under zero- and few-shot prompting. Under the illustrative cost model (c_ok, c_triage, c_wrong) = (0, 1, 5), TF-IDF with logistic regression and a threshold of 0.60 selected on calibration reduces test cost from 0.763 to 0.472, a 38.1% relative reduction (paired ratio-bootstrap 95% CI [35.7%, 40.4%]), while lowering wrong routes from 15.3% to 3.5%. Frozen calibration-selected policies also beat both always-route and always-defer on held-out test data in 14 of 16 settings in a post-hoc cost-parameter analysis. The strongest prompted configuration reaches 0.477 accuracy on a stratified 1,000-ticket test subset, versus 0.879 for fine-tuned DistilBERT on the same tickets. Isotonic lowers expected calibration error without reliably improving discrimination: GPT-4o-mini few-shot has one confidence value before and after recalibration and correctness AUROC 0.500. We therefore report correctness AUROC and tie-aware risk-coverage area alongside ECE. At the analytically derived break-even probability 0.80, seven prompted configurations defer all tickets; Haiku few-shot routes only 19/1,000. An LLM-only agreement control and margin/entropy rejection baselines further separate calibration from discrimination. The findings are scoped to this dataset, prompt design, model endpoints, and illustrative costs; the work is an evaluation framework, not a new abstention algorithm or universal deployment certification.

0.46

Post-Audit Mirage: Identifiability Limits of Offline Agent Verification

cs.AI cs.LG ·2026-10-22 ·Aman Behera, Payal Payal
publishedbenchmarktheorycode/data
Abstract · Accepted at the Workshop on Who Verifies the Agents? Toward Reliable Agent Development, NeurIPS 2026. 17 pages. Code: https://github.com/bei…

Offline tests can fail to reveal whether an agent update will remain safe after deployment. We operationalize this post-audit mirage using matched safe and harmful lifecycles that expose identical verifier-visible evidence while an isolated scorer records opposite deployment truth. Our theorem applies standard observational-equivalence and indistinguishability ideas to this executable audit boundary and states the resulting error-abstention tradeoff. The Post-Audit Mirage benchmark makes that boundary executable across authorization, scheduling, and shared-capacity environments. IdentifiedRangeMonitor uses fresh group-labelled harm outcomes to return deploy, hold, cannot determine, or unsupported. Experiments show that monitoring can recover the missing distinction in favorable settings but may remain undecided for rare groups, small effects, or missing observations. Verifier-agent evaluation must therefore test both calculation validity and whether the available evidence supports the deployment decision.

0.45

DISC: Domain-Integrated Similarity Cost for Time-Series Dataset

cs.LG ·2026-10-22 ·Yen-Ku Liu, Hongjie Chen, Franck Dernoncourt, Ryan Rossi
published
Abstract · Accepted as a conference paper at IEEE BigData 2026

The growth of time-series datasets and foundation models has made dataset comparison a central part of time-series workflows. Assessing source-target transfer before model adaptation requires similarity estimates that can summarize transfer-related dataset relationships without exhaustive source-target evaluation over every candidate pair. Many metrics have been proposed for measuring similarity between time-series datasets, but they capture different dataset relationships and vary in how well they align with downstream utility across tasks. We evaluate this gap by measuring the correlation between source-target similarity scores and downstream forecasting errors in out-of-distribution and transfer-learning settings. We propose Domain-Integrated Similarity Cost (DISC), a learnable aggregation framework that uses calibration transfer observations to select, weight, and combine similarity metrics for different downstream tasks. By integrating complementary metric values, DISC better distinguishes source-target dataset differences that matter for downstream behavior. After calibration, DISC scores source-target pairs using only their similarity-metric values and yields higher alignment than single-metric references on the reported split. These results show that time-series dataset similarity is task-dependent, and that learned metric integration provides a stronger proxy for downstream transfer behavior than manually selecting one similarity metric.

0.45

Post-Hoc Feature Selection Layer for Neural Networks Interpretability

cs.LG cs.AI ·2026-10-22 ·João Kenji Suwa, João Pedro Silveira e Silva, Bruno Iochins Grisci
published
Abstract · 6 pages, 3 figures, 4 tables. Accepted and published at the 2026 International Joint Conference on Neural Networks (IJCNN), Maastricht, Neth…

The interpretability of complex neural networks remains a critical challenge, especially for models already deployed in high-stakes domains. To address this, we adapted the Feature Selection Layer (FSL) to optimize feature weights post-training, a variant we term post-hoc FSL. Our approach reframes the FSL as a lightweight, trainable module that integrates with already frozen pre-trained models on tabular datasets to highlight the features the original model considers most important. This post-hoc FSL learns feature relevance by fine-tuning its weights based on the original model's learned outputs. Crucially, this process is non-invasive, operating without altering the original model's architecture or its learned parameters. We conducted our experiments using both statistical and visual metrics, including accuracy, F1 score, recall, precision, weighted t-SNE and silhouette score, and also analyzed the stability of the post-hoc FSL on high-dimensional synthetic and real-world tabular datasets. We compared the post-hoc FSL feature weighting method using these metrics against the original embedded FSL and other post-hoc interpretability methods, such as Integrated Gradients, Noise Tunnel, DeepLIFT, Gradient SHAP, and Feature Ablation. Experimental results demonstrate that the post-hoc FSL feature weighting method successfully identified relevant features across the different datasets, maintaining the predictive power of the original neural network while enhancing its interpretability.

0.45

Beyond Heuristics and Commercial Navigation: An AI-Augmented Decision-Support for Ambulance Travel Time Estimation

cs.LG ·2026-10-22 ·Muhammad Adil Abid, Johan Holmgren, Fabian Lorig, Jesper Petersson
publishedunusual
Abstract · 8 pages, 4 figures, 2 tables, accepted at CoopIS 2026

Accurate ambulance travel time estimation is critical for emergency medical dispatch, yet existing decision-support practices often rely on static lookup tables or simplified assumptions regarding how ambulance travel time differs from ordinary passenger-car travel. For instance, they assume ambulances simply travel a fixed percentage (e.g., 5%) faster than ordinary passenger cars. However, such assumptions are rarely validated using real ambulance operational data and may fail to capture the effects of trip priority, temporal context, and origin-destination. We address these limitations using historical ambulance trips from the Skåne region in Sweden. Our analysis reveals that ambulance travel times vary substantially by hour, day, and trip priority, and that commercial navigation systems fail to capture emergency-specific dynamics. We contribute a machine learning system to more accurately estimate ambulance travel time. To make the system interpretable, we explored SHapley Additive exPlanations and permutation importance techniques to ensure that our system captures operationally meaningful patterns.

0.45

Network topology shapes consensus and vulnerability in LLM populations

cs.AI physics.soc-phcs.CY ·2026-10-22 ·Roberto Passaro, Ariel Flint, Romualdo Pastor-Satorras, Andrea Baronchelli, Luca Maria Aiello
Abstract

As large language models (LLMs) are deployed in interacting populations, their collective behaviour depends not only on the models but also on who interacts with whom. Studies of convention formation in LLM populations have mainly tested well-mixed interactions. We place policies extracted from four LLMs on fully connected, random regular, and heterogeneous networks, with $10^3$ to $10^5$ agents playing the naming game. Topology affects how quickly consensus emerges, sometimes whether it emerges at all, and, in rare cases, which convention wins. Sparsest networks slow consensus up to twelvefold. Yet, by preserving correlations between persistent neighbours, they restore consensus in some cases that fail to converge under well-mixed interactions. Networks also govern vulnerability to adversarial attack. We measure the critical mass of committed agents needed to overturn an established convention. Tipping thresholds follow the same ordering across models and conditions: hubs $\leq$ random $\leq$ low-degree nodes. On the most heterogeneous topology, median critical masses are 3%, 16%, and 35%, respectively; targeting hubs requires one-sixth as many committed agents as random placement. For most models, networked populations are easier to tip than well-mixed ones, so well-mixed estimates can overstate robustness. Assessing the safety of multi-agent AI requires mapping who interacts with whom.

0.44

PriMa-Causa: Causal Prescriptive Maintenance

cs.LG cs.AIeess.SY ·2026-10-22 ·Felix Saretzky
Abstract · PhD thesis, University of Luxembourg, 2026. 203 pages. Also available at https://hdl.handle.net/10993/69383

Automated manufacturing requires data-driven decisions under uncertainty and sparse observational records, and unplanned downtime is a dominant cost of getting them wrong. Prescriptive maintenance recommends a maintenance action against an impending failure rather than merely predicting it. Established reference architectures couple predictive models with rule- or optimisation-based recommendation logic. Selecting a corrective action, however, requires an interventional quantity: how the line behaviour would change if the action were set. Manufacturing data is observational, so a model fitted to predict the outcome estimates an associational quantity and may misdirect the choice of action without the prediction error revealing it. Causal estimators target the interventional quantity but require an expert-led causal study per case, for which the bounded decision window of operation leaves no room. This dissertation analyses how causal reasoning can be embedded into prescriptive maintenance so that actions are ranked by their estimated interventional effect. The result is PriMa-Causa, built from three components: a causal reasoning layer that integrates the estimate into the PriMa reference model, a parametric simulator that supplies the interventional ground truth absent from manufacturing data, and a graph-conditioned tabular causal foundation model that, pre-trained once on this synthetic production-line prior, answers interventional queries in a single forward pass. PriMa-Causa attains a lower prescription error than associational baselines, with the largest margin under weak overlap and sparse context, and a partially disclosed causal graph reduces this error further. The causal inductive bias of the synthetic prior and the graph conditioning can thus replace per-case estimator fitting, a step toward automating causal inference for autonomous and self-healing maintenance.

0.44

Aligning Large Language Models with Human Feedback: Mathematical Foundations and Algorithm Design

cs.LG cs.CL ·2026-10-22 ·Siliang Zeng, Luca Viano, Chenliang Li, Jiaxiang Li, Volkan Cevher +4 more
published
Abstract

This article provides an introduction to the mathematical foundations and algorithmic frameworks used to align Large Language Models (LLMs) with human intentions, preferences, and values. We discuss standard alignment techniques, such as fine-tuning (SFT), reinforcement learning with human feedback (RLHF), and direct preference optimization (DPO). We also explore the theoretical underpinnings of learning from human preferences, drawing connections to inverse reinforcement learning (IRL) and discrete choice models. We present state-of-the-art algorithms in a tutorial style, discuss their advantages and limitations, and offer insights into practical implementation. Our exposition is intended to serve as a comprehensive resource for researchers and practitioners, providing both a foundational understanding of alignment methodologies and a framework for developing more robust and scalable alignment techniques.

0.44

Every Weight Is a Path of Shared Discrete Decisions

cs.LG ·2026-10-22 ·Ammar Ali, Stamatis Lefkimmiatis
community pick
Abstract

We introduce TWIG (Tree-structured Weights with Inherited Gates), a representation in which every weight of a neural network is the end of a path of small discrete decisions: a sign, a shift by a power of two, a zero gate, an index into a shared table. Weights that belong together share the upper part of their paths, and each decision is inherited by every weight below it, so it is stored once and applied once. This simple change has three consequences. Any trained network can be written as a TWIG, and when the decisions are powers of two its weight layers become circuits made almost entirely of wires and adders, suited to chips whose weights are fixed at fabrication. Pruning, quantization, low-rank factorization, codebooks and weight sharing become edits of one object, so a single optimizer can choose among all of them for a whole network. And choosing the decisions is the ground-state problem of a higher-order Boltzmann machine whose interactions follow the paths of the tree, the problem class that annealers and quantum optimizers are built to solve. Today these questions are handled apart: each compression method brings its own format and solver, a circuit with fixed weights needs a dedicated constant multiplier for every weight, and discrete solvers reach only narrow subproblems such as rounding. Weight formats have long been moving toward shared decisions, through group scales, shared exponents, microexponents and sparse weights; TWIG takes this trend to its end, and every common format is a shallow TWIG at no extra cost. On five weight matrices of a large language model, a flat tree whose signed power-of-two terms are chosen per weight for circuit cost, with no group scales, needs 9--15\% fewer additions than GPTQ at equal layer error at 3- and 4-bit accuracy on typical matrices, and 20--55\% fewer on matrices with strong outliers.

0.43

Evolving Federated Density Peak Clustering with Shuffler-Based Anonymization

cs.LG cs.CRcs.DCeess.SY ·2026-10-22 ·Miha Ožbot, Seiichi Ozawa, Igor Škrjanc
publishedbenchmarkcode/data
Abstract · 30 pages, 4 figures. Published open access (CC BY 4.0) in Evolving Systems 17, 123 (2026). Code: https://github.com/mihaozbot/eFDPC

In this paper, we present a novel one-shot federated clustering approach based on evolving fuzzy clustering and shuffler-based anonymization. In federated clustering, each round reveals additional model information to the server and can increase the opportunity to infer properties of local datasets, even when raw samples are never shared. Consequently, fewer client-server exchanges and unlinkability are important, so transmitted information cannot be attributed to any single client. Data-stream clustering algorithms, which learn in a single pass over the data, align well with the one-shot federated learning paradigm, where a single communication round can be sufficient. The proposed Evolving Federated Density Peak Clustering (eFDPC) method combines evolving fuzzy data-stream clustering and density peak clustering in a federated setting. Each client independently evolves micro-clusters from private data mini-batches through cluster addition and merging. To limit communication, density peak clustering is applied to the learned micro-clusters and only the selected peaks are transmitted, rather than the full local models. A shuffler-based layer is introduced between the clients and the server to anonymize contributions prior to reconstruction and global peak selection. We consider a peak-only variant, eFDPC-P, as well as eFDPC-G with server-side label propagation. Experiments on synthetic and real-world benchmark datasets show that eFDPC outperforms recent federated clustering baselines in one-shot settings and scales to higher-dimensional, non-IID data.

0.41

Uncertainty Quantification for Financial Foundation Models: A Survey

cs.LG q-fin.GN ·2026-10-22 ·Zijie Zhao, Mingjun Sun, Shenbo Xu
publishedsurvey
Abstract · Accepted at the 7th ACM International Conference on AI in Finance (ICAIF 2026). 8 pages

Financial foundation models increasingly support workflows spanning financial question answering, disclosure analysis, forecasting, and decision-making. In these settings, uncertainty may concern whether evidence is valid and available at the relevant time, whether an interpretation is justified, which outcomes remain plausible, or which action is preferred under costs and risk constraints. Because these objects arise at different points in a workflow, uncertainty cannot be reduced to confidence in a final output. We introduce a source-aware framework that distinguishes six sources of uncertainty across evidence and context, model and inference, and outcome and action, while treating agentic workflows as the execution layer through which uncertainty can propagate, change form, or be reduced. We develop a mechanism-based taxonomy grounded in how methods construct or verify uncertainty and reliability representations, and use it to assess their applicability across financial task domains. We further map finance-specific empirical studies across the taxonomy and task domains to characterize the current research landscape. Our analysis shows that most reviewed studies address uncertainty in individual parts of financial workflows, with limited work connecting evidence and reasoning reliability to downstream prediction and constrained action. By connecting where uncertainty arises, how it is represented, and how it bears on downstream decisions, this survey provides a unified foundation for studying and evaluating uncertainty in financial foundation models.

0.41

Scaling Laws for Safety Probes

cs.AI cs.CLcs.CRcs.LG ·2026-10-22 ·Yuchen Wu, Jerry Wei
code/datahot direction
Abstract · 62 pages, 23 figures, 33 tables

Safety probes trained on a model's activations are a standard tool for monitoring language models in real time. Each time a new model is released or a new harm area needs to be monitored, the probe must be retrained, which requires new labels and a choice of probe architecture. Yet neither the model nor the labels tell us in advance how well the resulting probe will perform. Finding out requires training and comparing many probe configurations. In this paper, we ask whether probe performance can instead be predicted in advance from model size and the number of labels. Existing safety datasets are insufficient for this, as they are either limited in scale or already saturated. We therefore construct six new safety datasets of sufficient size and difficulty. On these datasets, we train two probe architectures on 20 base models from five model families, with 32 to $10^{6}$ labels for each model. We measure performance using probe error, defined as 1 - AUROC, where lower values are better. We find that probe error follows two simple scaling laws. First, the lowest probe error a model can reach falls as a power law in its size. Second, probe error decreases towards its floor as the number of examples increases; the slope increases as the target model size increases. Both laws hold across tasks, model families and probe architectures, and only their fitted constants differ. When fitted on the smallest models of a family, the laws predict the error of the next larger model to within 0.02 in 72% of cases. They also identify whether more labels or a larger model is the cheaper way to reach a target error in 73% of cases. Finally, they choose the probe architecture with an extra error of only 0.006 on average when they pick the worse one. These laws allow safety researchers to determine how to create the best probe for any given model and label budget.

0.40

EpiSEM-Bench: Evaluating Safety and Evidence Recovery in LLM Agents under Missing Evidence

cs.AI cs.LG ·2026-10-22 ·Qijin Zhang, Hanzhen Wang, Yuehan Zhao, Haoyu Dai, Yao Shu +2 more
benchmark
Abstract

Tool-using language model agents rely on environmental facts to decide whether an action is permitted. When a tool response omits a safety-relevant fact, an agent may carry out a prohibited action even though the user request is benign and every returned statement is true. We introduce \bench{}, an executable benchmark for evaluating unsafe execution, safety boundaries, and evidence recovery in LLM agents under missing evidence. It contains 300 matched unsafe/safe task pairs across six domains. By systematically suppressing critical facts, the benchmark measures unsafe execution under critical evidence omission and determines the minimum observed failure cardinality, including failures with full evidence and cases with no observed failure after exhaustive testing. An independent audit tool tests whether agents seek missing facts and then withhold prohibited actions while completing permissible tasks. The benchmark reveals substantial variation across the ten models in omission robustness, evidence acquisition, and appropriate action. Resistance to missing evidence does not necessarily translate into useful recovery. Recovery access reduces unsafe execution for all ten models, but agents differ in whether they actively obtain the missing facts and use them to act appropriately in both worlds.

0.40

Towards Trustworthy AI-Enabled Emergency Medical Dispatch: Uncertainty-Aware Ambulance Travel Time Estimation

cs.LG ·2026-10-22 ·Muhammad Adil Abid, Johan Holmgren, Fabian Lorig, Jesper Petersson
published
Abstract · 15 pages, 5 figures, 5 tables. Accepted at ICSOB 2026

Emergency medical dispatch increasingly relies on AI-enabled decision support systems when making ambulance dispatch decisions. In this process, accurate ambulance travel time estimates are essential for timely resource allocation and prehospital care. However, existing ambulance travel time estimation approaches typically produce only single point predictions. Consequently, they do not indicate how reliable an individual estimate is or how much uncertainty it carries. This limitation is particularly problematic in safety-critical emergency medical service dispatch operations, where trips with similar expected travel times may exhibit substantially different levels of uncertainty. Furthermore, real-world ambulance trip logs may contain domain-specific artefacts, such as co-located healthcare facilities and physically implausible travel speeds. These artefacts can degrade point prediction accuracy and affect the prediction task on which subsequent uncertainty estimation depends. To address these challenges, we propose an uncertainty-aware methodology for ambulance travel time estimation and evaluate it through a retrospective real-world scenario study using two years of ambulance trip data from the Skåne region in Sweden. The proposed methodology moves beyond average-case point predictions by generating context-adaptive prediction intervals with stable empirical coverage across the evaluated operationally important contexts. The resulting prediction intervals provide explicit information about predictive uncertainty that can support more transparent and risk-sensitive ambulance dispatch decision-making.

0.40

Future of Time Series Anomaly Detection: From Historical Review to Unified Outlook

cs.LG cs.AI ·2026-10-22 ·Zhijie Zhong, Zhiwen Yu, Pengyu Li, Liujian Zhang, Qiaoli Ni +4 more
top venuesurveybenchmarkcode/data
Abstract · 18 pages, 5 figures, 4 tables. Accepted by IEEE Transactions on Artificial Intelligence. DOI: 10.1109/TAI.2026.3746935

Time series anomaly detection (TSAD) is a foundational problem in data mining, aiming to identify observations that deviate significantly from nominal temporal patterns. Driven by escalating data volumes and application complexity, recent research has yielded diverse architectures, including diffusion models, state space models, and large language models (LLMs). However, existing surveys provide limited coverage of these emerging designs and lack a unified treatment of multivariate anomalies, cross-domain unified modeling, metric reliability, and contemporary evaluation challenges. To bridge these gaps, this survey delivers three core contributions: (1) We establish a refined anomaly categorization and a structured taxonomy of paradigms reflecting modern advances, comparing detection mechanisms and charting developmental trajectories across paradigms. (2) We provide an extensive analysis of evaluation methodologies and conduct comprehensive experiments to assess metric fidelity, complemented by an open-source benchmark repository encompassing algorithms, datasets, and evaluation tools. (3) We outline key open challenges, offering forward-looking perspectives on dynamic thresholding, model interpretability, zero-shot transfer, multi-modal alignment, and foundational models for time series. By addressing existing theoretical and empirical limitations, this survey establishes a coherent framework to guide future TSAD research and practical deployments.

0.39

Governed Self-Evolution for Autonomous Investing: A Multi-Agent LLM System with an Invariant Core

cs.AI cs.CLcs.MAq-fin.PM ·2026-10-22 ·Yebo Feng, Jingyang Dai, Liangrui Pan, Nelson Lau, Jiahua Xu +1 more
code/data
Abstract

LLM agents that act on live Web data must combine heterogeneous evidence, remember what they have done and improve from outcomes without weakening their own safeguards. We study this problem in autonomous investing, where every action is irreversible and based on untrusted Web content. We present Steward, a multi-agent LLM system that draws the boundary of self-evolution in its type system. First, nine role-specialised agents decide jointly: four analysts read news feeds, SEC filings, macroeconomic series and market data and emit typed reports, from which a Portfolio Manager sets target positions that a protected Risk Manager audits. Second, a seven-layer shared memory, assembled by code from the system's own record, gives each decision its position context, the market narrative, calibration and the agent's past decisions. Third, the system improves itself by review: code grades every trade and forecast nightly and writes the grades into memory, and each quarter a reflector proposes prompt learnings that a deterministic gate applies reversibly and replay of recorded inputs verifies. All three operate around an invariant core of order arithmetic, risk rules and discipline-agent prompts that no self-edit can reach. In a five-month live deployment (102 trading days, 1,632 LLM calls), the system contained a 24.6% sector drawdown to a 0.3% loss; in replay of its recorded inputs, a single model given the same raw data traded more and lost 1.35 points against the market over five days where the joint design showed no significant loss, and no injected instruction was followed; and a 2015-2026 backtest shows the core alone lowering the maximum drawdown under all 50 random entry policies. The reflector traced the dominant failure to its root cause, and the gate applied a learning that the agent adopted on trigger evenings and on no control evening. Code: https://github.com/yebof/quant-agent

0.37

Random Initialization as a Baseline for Time Series Foundation Models

cs.LG ·2026-10-22 ·Pinar Sungu Isiacik, Georgiana Ifrim
published
Abstract · Accepted by the NeurIPS-26 Workshop on TS-LIMITS

Time series foundation models (TSFMs) use self-supervised pretraining to learn reusable representations for downstream tasks such as multivariate time series classification (MTSC). We examine how much pretrained weights contribute beyond the architecture itself by comparing seven publicly released instances from four TSFM families with architecture-matched randomly initialized counterparts on the UEA MTSC archive. For each instance, five random initializations are evaluated using four classifiers on frozen representations: ridge classification, logistic regression, RBF SVM, and random forest. After model-specific exclusions due to computational constraints, the study comprises 183 paired dataset-level comparisons. Pretraining improves mean accuracy for all seven instances, but the gains are modest, ranging from 1.43 to 5.19 percentage points, with an overall average improvement of 2.58 points. Three of seven per-instance Wilcoxon signed-rank tests have raw $p$-values below 0.05, but only NuTime remains significant after Holm correction. A secondary pooled test across all 183 pairs gives $p < 0.001$, while the corrected per-instance tests remain the primary evidence. We also compare with ROCKET, Hydra, and Quant, which remain competitive without requiring large-scale pretraining. These findings suggest that MTSC evaluations should distinguish the effect of pretrained weights from that of the underlying architecture and should include both architecture-matched random initialization and strong non-pretrained baselines.

0.37

AgenticEvolve: Evolutionary Program Search with Persistent Language Model Agents that Design and Run Experiments

cs.NE cs.AIcs.LG ·2026-10-22 ·Ze-Yu Peng, Qing-Yu Lan, Zhan-He Wang, Jun Zhang, Yun-Song Piao
code/data
Abstract · 23 pages, including supplementary material. The source code is available at https://github.com/pengzy-cosmo/AgenticEvolve

Evolutionary program search with large language models keeps the structure of an evolutionary algorithm and uses the model as its variation operator. Typically, the model generates each candidate program in a single request from parent programs and their scores. We present AgenticEvolve, an evolutionary program search framework in which each island is run by a persistent large language model agent. Between evaluations, the agent designs and runs experiments on the candidate program. To measure the contribution of experimentation, we compare AgenticEvolve with No-analysis, an otherwise identical configuration with experiments disabled. On two heuristic design tasks with three large language models, AgenticEvolve produces better programs than No-analysis and OpenEvolve on both hidden and unseen instances. It reaches the baselines' final quality with far fewer evaluations and comparable cost. On twelve equation discovery problems, it recovers the ground-truth structure of more equations than No-analysis and existing language-model and genetic programming baselines.

0.36

Dual Exam: Controlled Evaluation of Multimodal Agents and World Models

cs.AI ·2026-10-22 ·Zhefan Rao, Meng Chu, Xuan Billy Zhang, Jingfan Chen, Guohao Cai +87 more
community pick
Abstract

World Models (WMs) can support Multimodal Agents (MAs) through simulated experience, while Agent trajectories can inform World Model refinement. Evaluating this interaction requires separate evidence about action selection and state evolution. We introduce Dual Exam, which tests Agents under fixed environment response mechanisms and World Models under fixed action interventions across physical, digital, social, and scientific domains. The Agent track compares 10 configurations on 1,088 items in 15 reporting categories. The World Model track contains 400 cases across eight scenarios; 15 configurations have full coverage and one covers six scenarios. GPT 6 Astra leads the case-weighted Agent Index at 31.3, and GPT-Image-2 leads the full-coverage WM Overall at 21.7. These scores use different tasks and metrics and are not directly comparable. A 150-case pilot of one coupled pairing records zero task successes and motivates component-level diagnosis. Results reveal uneven task performance and failures to preserve task-relevant state. Their value for predicting coupled success or sustained reciprocal learning remains to be established.

0.36

nnInteractive Agent: General-Purpose AI Agents as 3D Image Annotators

cs.LG ·2026-10-22 ·Fabian Isensee, Constantin Ulrich Harsy, Maximilian Rokuss, Benjamin Hamm, Philipp Vollmuth +1 more
benchmarkcode/data
Abstract

Interactive 3D segmentation models such as nnInteractive turn a few spatial prompts (points, scribbles, boxes) into accurate volumetric masks, but a human still does the annotation work: finding each structure, placing prompts, checking the result and correcting it. General-purpose AI agents can now interpret images, call tools and write code. We show that such agents can take over this work without additional training: the user makes a request in plain language, and the agent works out what has to be segmented, locates the targets, prompts nnInteractive, inspects the 3D result, corrects it and responds to feedback. Asked for 'the muscles used for chewing', an agent identifies and segments the masseters and pterygoids; asked to count liver lesions, it writes its own code to find candidates and splits a merged lesion on request. Our open-source nninteractive-agent provides nnInteractive as tools over the Model Context Protocol (MCP), together with a skill that encodes how experienced users prompt it. On a benchmark spanning CT, MRI, PET and micro-CT of humans, a mouse and an ant (14 images; 51 targets from organs and tumours to muscles, foot bones, vessel trees and an insect brain), Claude Code and Codex reach a mean Dice of 0.81 and 0.84 from the request alone; when told where each target is, both reach 0.85 and prompt nearly as effectively as an oracle that places prompts using the reference segmentation. A text-prompted 3D model (VoxTell) is more accurate on most standard radiological targets but less robust: it returns an empty mask for one of them and fails on the leg muscles, the foot bones, the mouse and the ant, where the agents remain accurate. Code: https://github.com/MIC-DKFZ/nninteractive-agent

0.36

Nested Sparsity for Scalable Pretraining and Continual Learning

cs.LG ·2026-10-22 ·Tianzhe Chu, Druv Pai, Yi Ma
code/data
Abstract · 10 pages main body

Mixture-of-experts (MoE) transformers, the standard backbone for modern large-scale deep learning, employ conditional computation: only a sparse subset of model weights is used in each forward pass. In MoE transformers, this sparsity arises in the MoE operator, which generalizes the multi-layer perceptron in the classical transformer. In this work, we explore an alternative mechanism for conditional computation that analogously generalizes the attention operator. The sparsity induced by this mechanism naturally nests (i.e., hierarchically composes) with the sparsity induced by the MoE operator. We show that a transformer employing this nested sparsity mechanism, which we call Nestor, performs better than a conventional MoE transformer at equal compute and equal sparsity in language model pretraining. We also show that Nestor is well-suited for continual pretraining: it can be adapted to longer contexts more effectively than conventional MoEs, as well as adapt more efficiently to new domains while forgetting less on old domains. Furthermore, we demonstrate that these benefits persist as model size, dataset size, and context length grow. Our results demonstrate the potential of alternative conditional computation mechanisms for deep neural network architectures. Code and checkpoints are available on GitHub at https://github.com/Ma-Lab-Berkeley/Nestor .

0.36

Diagnosing Personalization Failures in Mobile GUI Agents

cs.AI ·2026-10-22 ·Zhixin Lin, Dongliang Xu, Jungang Li, Shidong Pan, Jun Feng +4 more
code/data
Abstract · 24 pages, 8 figures

Unlike conventional GUI agents, personalized GUI agents require user preferences to guide sequential execution over changing GUI states. This raises two challenges: determining HOW each preference should alter execution strategy and WHEN it should take effect along the interaction trajectory. Existing benchmarks primarily evaluate overall task success, offering limited insight into these questions. We introduce PrefDiag-Bench with two components: (1) a categorization of preferences into five types, characterizing HOW they affect execution; and (2) a stage-wise diagnostic framework with 1,664 probes, tracing WHEN preferences should take effect and where failures occur. We evaluate eleven GUI agent systems, including end-to-end agents and agentic workflows, under two forms of user context: explicit user profiles and behavioral memory logs. Although existing agents achieve relatively high completion rates on non-personalized tasks, performance is substantially lower on preference-personalized tasks. Even the best-performing model, Seed-2.0-Pro, achieves only 47.17% on Process preferences under explicit user profiles. Agents struggle particularly with Ranking and Process preferences. Further analysis reveals two key capability bottlenecks: selecting relevant evidence from user history and recognizing when preferences should take effect during execution. Controlled interventions that explicitly provide relevant evidence improve end-to-end task success by up to 7.06%. PrefDiag-Bench offers a more systematic evaluation of GUI agents' personalization capabilities beyond overall task success. Code and data are available at https://github.com/Zhixin-L/PrefDiag-Bench.

0.36

FairFRESH: A Neuro-Symbolic Framework for Fair and Explainable Household Food Recommendation

cs.AI cs.GT ·2026-10-22 ·Kodur Sai Vinay Sathvik, Kakoli Bora, Tejaswini R. M
publishedtheory
Abstract · 22 pages, 4 figures. Accepted and presented at ICCIS 2026; this version includes additional theoretical analysis and experiments

Recommender systems typically serve one user at a time. However, in households, members share a finite pantry and follow different dietary rules, making independent recommendations prone to resource conflicts and dietary violations. We formulate household food recommendation as constrained fair division and present FairFRESH, a neuro-symbolic framework that learns per-member utilities, allocates foods under explicit welfare objectives, enforces dietary rules through a deterministic filter, and explains recommendations using Shapley and Banzhaf attributions. Theoretically, we show that no finite penalty guarantees compliance near a dietary-rule threshold, establish approximation guarantees for greedy allocation under utilitarian and smoothed Nash objectives in the stock-free model before filtering, and characterize the scale invariance of Nash welfare. We also use the minimax theorem to analyze rotating shared meals. On NHANES-derived data, the learned utility model achieves Recall@20 of 0.293, outperforming five learned recommenders (at most 0.244) but not a previous-day repetition heuristic (0.321). On held-out four-person households, Nash allocation reduces Gini disparity by 42% relative to conflict resolution while improving Recall@20 from 0.180 to 0.197. It comes within 2% of an upper bound on the smoothed Nash optimum, and rescaling one member's utilities changes at most 1.5% of assignments. All recommendations satisfy the encoded dietary rules, while attribution scores correlate with utility loss upon component removal (r = 0.76). In simulation, shared meals reduce pantry use by about one-third, and weekly rotation increases the worst-off member's utility by 38% over the best fixed meal.

0.36

Learning to Be Monitored: Causal Bounds and Intervention-Aware Post-Training Recipes for Chain-of-Thought Monitorability

cs.LG cs.AIstat.ML ·2026-10-22 ·Qiang Fu, Xinyue Li, Andre Wibisono, Arman Cohan
theorycode/datahot direction
Abstract

Chain-of-thought (CoT) monitoring aims to detect potentially harmful behaviors in large language models (LLMs) from their CoTs. Intervention-based evaluations measure this property by asking whether a monitor can detect the role of an intervention from the CoT when the targeted behavior occurs, typically using minimal-criterion sensitivity ($\mathsf{TPR_{mc}}$) proposed by~\citep{guan2025monitoring}. However, a target behavior observed under intervention is not necessarily caused by it, and we show that this metric can overestimate how often a monitor detects target behaviors actually caused by the intervention. Specifically, we establish $\mathsf{TPR_{mc}}$ as an upper bound on this causal detection rate and derive a complementary lower bound, which provides a tighter evaluation framework for CoT monitorability. With this framework, we study whether LLMs can be trained to make their CoTs easier to monitor. We first construct a training dataset of paired control--intervention tasks spanning multiple intervention types. Using the dataset, we design and apply three intervention-specific training methods based on off-policy self-distillation, reinforcement learning, and on-policy self-distillation to train separate variants of each base model. Across various model-monitor configurations, all three methods generally improve CoT monitorability over the corresponding base models under our extended evaluation framework. Our code is available at https://github.com/xinyue-lxy/learning-to-be-monitored.

0.35

The Delegation Divide: An Open Index of AI-Agent Delegability across Occupations and Countries

cs.AI econ.GNcs.CY ·2026-10-22 ·Cristian Espinal Maya, Santiago Jiménez Londoño
code/data
Abstract · 32 pages, 6 figures, 4 tables. Preregistered at OSF: 10.17605/OSF.IO/9ZNTF. Data and code: doi:10.5281/zenodo.23552918; repository and MCP s…

35.2% of task time in world employment is at least partly delegable to AI agents in 2026, and only 2.5% can be handed over in full. Existing measures score exposure to generative AI, mostly for the United States. We build the Agent Delegability Index (ADI): two language models rate all 18,796 O*NET 30.2 task statements on a frozen, preregistered rubric of delegation levels and barriers; tasks are mapped to 433 ISCO-08 unit groups and weighted by ILOSTAT employment in 146 countries. Four findings. First, delegable employment rises with income, from 22.5% in low-income to 48.7% in high-income countries, a slope of 7.23 percentage points per log point of GDP per capita. This delegation divide is compositional: the physical barrier dominates the jobs poorer countries employ most. Second, at the occupation level ADI tracks exposure closely (r = 0.93 with Eloundou et al. and 0.89 with Felten et al.), above the preregistered bound of 0.80; ADI adds delegation levels and barriers on top of exposure, for the world, not a new ranking of occupations. Third, full delegation is rare and peaks for tasks that take a human about twelve minutes (11.8 minutes). Fourth, the preregistered adoption gap is not supported: across 101 countries, delegability and observed AI usage correlate at r = 0.86, and the gap between them is not wider in middle-income countries (p = 0.27). Model agreement is kappa = 0.77; a blind human rating of 300 tasks agrees with the model consensus at kappa = 0.85, and an agentic panel with web search at kappa = 0.87. We release the index, the atlas, the prompts and an MCP server at https://github.com/Cespial/delegation-divide.

0.35

A Fix for Softmax and Cross-Entropy

cs.LG ·2026-10-22 ·Jonathan T. Barron, Dan Gnanapragasam, Michael Tschannen
highly cited
Abstract

Standard implementations of softmax, log-softmax, and cross-entropy have two numerical bugs. The first bug, Softmax Collapse, rounds the target logit's gradient to zero if that one logit is much larger than the others, resulting in gradients that no longer sum to zero. The second bug, Softmax Cancellation, occurs when student logits are close to teacher logits, and produces non-zero gradients even when the student and teacher logits are equal. We present simple implementations of log-softmax, softmax, and cross-entropy that fix both bugs. Our fix eliminates the previously observed "slingshot" phenomenon, speeds up grokking with weight decay, prevents accuracy collapse without weight decay, stops student networks from drifting away from the teacher during distillation, and shows that after memorization, logit growth is bounded by an equilibrium between weight decay and Adam's $ε$ hyperparameter.

0.35

Artificial Intelligence in Mechanical Engineering: Foundations, Opportunities, Challenges, and Responsibilities

cs.AI cs.CEcs.LGstat.AP ·2026-10-22 ·A. Cunha, C. Nataraj, L. M. Lorenzo, H. Panetto, A. Scotti +1 more
published
Abstract

Artificial intelligence (AI) is reshaping mechanical engineering (ME), from modeling, simulation, and optimization to diagnostics, design, and technical communication. Drawing on a round-table at the 28th ABCM International Congress on Mechanical Engineering (COBEM 2025), this article presents a structured perspective for the Brazilian and international ME communities. Rather than an exhaustive survey, it examines how AI can enhance engineering practice without compromising scientific rigor. We distinguish statistical learning from human cognition and discuss the capabilities, limitations, and emerging agentic forms of AI. We then consider industrial adoption, workforce changes, and AI-assisted scientific discovery, alongside the energy and infrastructure demands of AI computation and their implications for sustainability and thermal management. The discussion emphasizes the integration of physical knowledge, structured reasoning, and domain expertise into AI systems, particularly in safety-relevant applications. Topics include physics-informed and hybrid learning, physically constrained modeling, validation, uncertainty quantification, and certification. We also examine educational, epistemological, and ethical challenges, including curriculum reform, human-agent collaboration, professional accountability, and the preservation of engineering knowledge. We conclude with principles for responsible AI adoption in ME: physical consistency, rigorous validation, explicit treatment of uncertainty, energy sustainability, human oversight, and ethical responsibility. AI should complement, not replace, engineering judgment.

0.35

Bifurcations from Weight Space: JEPAWG on Parametric Dynamical Systems

cs.LG cs.AInlin.CD ·2026-10-22 ·Tobias Göbel, Miranda C. N. Cheng
published
Abstract · 13+9 pages, 5+8 figures, 3+4 tables. A version of this work appears at the NeurIPS 2026 Workshop on Neural Network Artifacts as a New Data M…

Trained network weights are increasingly treated as data, but there is rarely ground truth about what a population of models should encode, which hampers systematic validation of weight-space methods. This makes parametric dynamical systems a natural test-bed: the dynamics change qualitatively at submanifolds of parameter space that are, in well-studied families, known analytically or numerically. Fine-tuning a forecasting network at each point of a parameter grid yields a model zoo indexed by a physical parameter, against which weight-space diagnostics can be checked. We build such model zoos for four dynamical systems and analyse them with JEPAWG, a recently proposed joint-embedding hypernetwork built around a learned parameter--weight representation. It was first introduced for neural samplers for lattice field theories, where its latent geometry located the phase transition and estimated the 2D Ising critical exponent. Applied to dynamical systems, JEPAWG is competitive on forecasting with a hypernetwork trained directly for that task, despite being supervised on the parameter--weight correspondence rather than on forecasts. Beyond forecasting, the norm of the parameter-derivative of the decoded weights locates Hopf points and curves, tracks a period-doubling cascade closely enough to give a qualitative estimate of the Feigenbaum constant, and detects a global bifurcation, which has no eigenvalue signature and is therefore invisible to linearisation. As such, JEPAWG provides interpretable insight into the dynamics that is not available from other hypernetworks. Together, these results indicate that JEPAWG is a workable instrument for reading physical structure off a population of trained networks.

0.34

How Firm Are AI Models' Ethical Boundaries? A Pay-Stub Editing Test of 15 Models

cs.AI cs.CL ·2026-10-22 ·Simiao Ren, Xingyu Shen, Tommy Duong, Tsang Ng, Ankit Raj
Abstract · 11 pages: 4 pages of main content, 1 page of references, and 6 appendix pages; 7 figures and 4 tables; ancillary data and code included

Can a small change in wording move an AI model from refusing a sensitive request to acting on it? We test 15 models on the same request to increase the income shown on a pay stub. Across 445 observable trials, models issue editing calls in 14.1% of cases when a loan purpose is stated, 33.1% when the purpose is omitted, and 66.2% when the change is described as fixing a printing error. Astra refuses all ten loan-purpose requests but issues calls in all ten purpose-omitted trials. Fable, Opus, and Muse Spark issue none across all three wordings; Sonnet issues some under the latter two. Response excerpts illustrate different ways of refusing, redirecting, or justifying an edit. The file is a marked test document and all tools are simulated. These results probe how boundaries behave in one setting; they do not prove an overall ethics ranking or that a harmful real-world edit succeeded.

0.34

Lean Finder V2: Semantic Search for Mathlib That Disambiguates User Queries

cs.LG cs.IR ·2026-10-22 ·Jialin Lu, Xiaoyu Liu, Hengrui Luo, Rodrigo Stehling, Linxi Liu +3 more
benchmarkpositiontheory
Abstract

We introduce Lean Finder V2, a semantic search framework for Mathlib that disambiguates user queries through complementary model-level adversarial training and input-level query rewriting. Human mathematicians and coding agents rarely search Mathlib by providing detailed context. In practice, they query short properties, notation, aliases, or hypotheses, leaving much of the intended mathematics implicit. In contrast, existing Mathlib search engines unrealistically assume context-rich queries. The scarcity of real-world short queries paired with ground-truth Mathlib statements makes this ambiguity even more intractable. We argue that this is essentially a data problem. Through adversarial training, we scale up the generation of labeled short and ambiguous queries, and learn to disambiguate user queries at both model-level fine-tuning and input-level query rewriting. On the human-curated MathlibQR benchmark, our Lean Finder V2 sets a new state of the art, with at least 11% relative improvement over the previous best Mathlib semantic search engines. On top of our state-of-the-art retriever, optimized query rewriting further improves MathlibQR retrieval. On unlabeled queries from human and coding-agent interactions, Lean Finder V2 is ranked as the strongest system. Integrated with Claude Code, Lean Finder V2 improves proof success on downstream theorem-proving task. Lean Finder V2 is available at: https://leanfinder.github.io/

0.34

Prospect Theory in Large Language Models: A Meta-Analysis of Loss Aversion, Diminishing Sensitivity, and Probability Weighting

cs.AI cs.CLecon.GN ·2026-10-22 ·Adi Tarabeih
Abstract

Prospect theory splits risky choice into three parts: loss aversion, measured by the relative weight of losses to gains, $λ$ (e.g., $λ= 2$ implies that losses weigh twice as much as gains), diminishing sensitivity to the size of outcomes (i.e., the curvature and linearity of the value function), and the nonlinear weighting of probabilities. I examined whether Large Language Models (LLMs) display the same properties as people by meta-analyzing 102 fitted estimates from 11 studies covering 32 language models. The three components emerge differently. Loss aversion is modest: the pooled $λ$ is 1.26 (95% CI 1.06 to 1.46), far below the strongest human meta-analytic estimate ($λ= 1.96$) and compatible with more conservative ones (1.07 and 1.31). Matched by design, LLMs are compatible with people when gains and losses are symmetric ($λ= 1.15$ against 1.07), but well below them when losses are smaller than gains (1.46 against 2.3), where human loss aversion is strongest. Value functions are close to linear rather than concave under the Tversky-Kahneman method (1.01, against a human value of 0.67), but more concave than in people under the Tanaka method (0.52, against 0.67), so diminishing sensitivity depends on how it is measured. Probability weighting shows a human-like inverse S shape ($γ= 0.81$ under Tversky-Kahneman and 0.84 under the Tanaka method), though on few studies. In sum, LLMs show little loss aversion and little diminishing sensitivity, but retain human-like probability distortion, yet these generalizations shift when looking at specific model families and fitting methods.

0.34

PointerBench: A Diagnostic Benchmark for Fine-Grained GUI Grounding

cs.AI cs.HC ·2026-10-22 ·Valentino Sacco, Saurabh Pandey, Alexander Florian Jercher, Maximilian Schilling
publishedbenchmarkcode/data
Abstract · Accepted at the 2nd Workshop on Agentic AI Benchmarks and Applications for Enterprise Tasks (AABA4ET), NeurIPS 2026

Multimodal Large Language Models have shown promising performance in computer-use tasks, from planning user workflows to grounding natural-language actions to precise UI targets. Existing end-to-end benchmarks primarily emphasize long-horizon task completion, while GUI grounding benchmarks remain focused on conventional controls and a limited set of professional interfaces. We introduce POINTERBENCH, a diagnostic suite of 1,500 screenshot-instruction pairs probing three complementary capabilities: thin sub-element identification and geometric grounding in spreadsheets (POINTERBENCH-Sheets), character-level grounding in long-text documents (POINTERBENCH-Text), and mixed GUI grounding across 100 professional applications (POINTERBENCH-Pro). We carefully curate each task with pixel-exact target geometry and evaluate both general-purpose multimodal models and 39 specialist GUI grounding configurations. We find text grounding to be the most challenging setting, with Claude Fable 5, GPT-5.6 Sol, and the leading native specialist UI-Venus-2-9B achieving 69.6%, 53.4%, and 25.8% success on POINTERBENCH-Text despite respective 83.2%, 77.7%, and 54.5% average success across POINTERBENCH. These results reveal substantial headroom in fine-grained GUI grounding across frontier and specialist models. We publicly release the benchmark and evaluation code at https://github.com/warmwindOS/pointerbench.

0.34

ADAPT for Science: A Multi-Agent System for Self-Evolving Scientific Optimization

cs.AI cs.MAphysics.optics ·2026-10-22 ·Zhaoran Feng, Yuchen Ni, Jianan Liu, Lemao Liu, Dejing Dou +1 more
Abstract · Includes supplementary material

Scientific discovery often requires identifying a small number of high-performing designs in vast, constrained search spaces. This is difficult when evaluations are costly, feasible solutions are sparse, parameters are strongly coupled, and physical constraints are strict. Recent AI for AI methods can automatically improve algorithms and optimization procedures. However, general purpose approaches are not designed around scientific priors, physical constraints, and limited evaluation budgets. We introduce ADAPT for Science, a multi-agent framework that treats scientific optimization as two coupled decisions: where to search and how to search. ADAPT uses historical designs to identify promising and diverse initial conditions, progressively concentrates computation on favorable search directions, and evolves the optimization procedure from execution feedback. We evaluate ADAPT on the design of gravitational wave detectors using a differentiable physics simulator and eight unseen detector topologies. Under matched scientific evaluation budgets, ADAPT achieves a mean loss of 0.204 +/- 0.101, reducing loss by 54.9% relative to the best classical optimizer and by 62.2% relative to the best general purpose agentic evolution baseline. ADAPT achieves the lowest loss on all unseen topologies among the evaluated baselines. ADAPT provides a step toward scientific optimization systems that continuously learn from prior experience and their own search outcomes.

0.34

RIDAR: Safeguarding Research Integrity in Autonomous Research Agents via Automated Detection and Runtime Intervention

cs.AI ·2026-10-22 ·Jiening Siow, Amy Xin, Junjie Wang, Ziyu Tang, Zijun Yao +5 more
benchmark
Abstract

Autonomous research agents are increasingly used to automate large portions of scientific research, yet they are prone to research integrity violations (RIVs): behaviors that compromise the validity, reproducibility, or faithfulness of the research process. We present RIDAR, a framework for automated RIV detection and runtime intervention. First, we introduce RIVT, a comprehensive taxonomy of 14 RIV types spanning the full research lifecycle. Building on RIVT, we present RIDAR-post-hoc, a framework to detect RIVs within agent trajectories. RIDAR-post-hoc achieves 85.2% F1 performance on RIV-Gold, a benchmark for RIV detection with 60 expert-annotated trajectories. Applying RIDAR-post-hoc to RIV-Wild, a dataset of 300 agent trajectories across 50 distinct research tasks, we detect 3,964 RIV instances in 99.7% of trajectories, revealing the prevalence of RIVs in current agents. Finally, we present RIDAR-runtime, a framework for dynamically detecting and correcting RIVs during agent runtime. RIDAR-runtime effectively reduces agent RIVs remaining at task completion by 36.8%. We will release our code and data to facilitate future research.

0.34

Evaluating Type-1 Diabetes Digital Twins by Decision Quality

cs.LG ·2026-10-22 ·Victor Li, Owen Tucker, Michael S. Hughes, Temiloluwa Prioleau, Shengpu Tang
code/data
Abstract · 12 pages, 1 figure, 7 tables. Code: https://github.com/Veeeeeee7/T1D-Digital-Twin-Benchmark

A digital twin (DT) for type 1 diabetes (T1D) can predict how an individual's blood glucose will respond to different insulin treatments without actually trying them. For each treatment, downstream statistics of predicted glucose trajectories, such as glycemic risk, can be computed to guide clinical decisions. However, the quality of DTs is usually measured by the accuracy of their predicted glucose trajectories, which may not reflect the quality of the downstream treatment decisions they inform. In this work, we directly evaluate DTs by this decision quality on downstream treatment ranking, i.e., whether a DT predicts glucose trajectories that can be used to rank candidate treatments as the real patient would. We present an evaluation protocol over 30 virtual patients from the FDA-accepted UVA/Padova model, each simulated under 21 different basal-bolus treatments. For each patient, a DT is fitted under one treatment and then produces glucose trajectories for the remaining 20 treatments. Our results show that prediction accuracy and decision quality disagree. A baseline that models the average patient, never seeing the individual's glucose data, has the lowest prediction accuracy of all six methods (RMSE: 63.6 mg/dL) but ranks treatments with the second-highest mean Spearman (0.766), while a black-box neural network that predicts glucose nearly twice as accurately (RMSE: 32.9 mg/dL) ranks them no better than a random ordering of the 20 treatments (0.049 Spearman). Therefore, selecting a DT based on prediction accuracy alone can prefer one that recommends treatments at chance, and decision quality must be evaluated directly rather than assumed to follow from prediction accuracy.

0.34

Zero-Shot Forecasting of In-Situ Soil Moisture and Temperature with a Time-Series Foundation Model

cs.AI ·2026-10-22 ·Mohammad Talaei, Mohammad Goudarzi, Maille Todd, Vanessa N. L. Wong, Iadine Chadès
published
Abstract · Accepted at the NeurIPS 2026 Workshop on Tackling Climate Change with Machine Learning

Soil moisture and temperature govern plant-water availability and microbial activity, making them key indicators for evaluating land-restoration efforts on degraded landscapes. However, the physical and data-driven models normally used to forecast them both need information that a newly instrumented restoration site does not yet have: hydraulic parameters for the former and months of per-site training data for the latter. We evaluate TimesFM 1, a 500M-parameter time-series foundation model, zero-shot on 15-minute telemetry from 27 sensor stations across three land-management regimes (Control, Treatment, Scald) at a semi-arid restoration site, forecasting 5 cm and 30 cm soil moisture and enclosure temperature at seven lead times from 1 h to 168 h. Using gap-aware rolling-origin evaluation across sensors, we find that TimesFM beats the best rule-based statistical forecasting baselines (persistence, seasonal-naive, or 7-day climatology) by 21-76% at sub-daily lead times but loses by 6-79% beyond a day, a pattern explained by a diurnal-alignment mechanism that holds for both soil-moisture depths and temperature. A CatBoost baseline trained walk-forward on the same origins with full history and weather covariates never wins overall, and we show that its weakness is partly a cold-start effect specific to short deployments. We additionally quantify the inference cost of zero-shot forecasting on commodity CPU hardware. Our results highlight a practical trade-off between zero-shot forecasting capability, predictive performance, and computational efficiency.

0.33

Not All Retained Clients Are Equal: Client Selection for Federated Unlearning

cs.LG cs.DCcs.ET ·2026-10-22 ·Thanh Linh Nguyen, Quoc-Viet Pham
top venue
Abstract · Accepted at NewInML @ NeurIPS 2026

Federated unlearning (FU) aims to remove a designated target client's influence from a trained model without the prohibitive cost of retraining from scratch. Existing methods largely treat unlearning as a post-hoc, stateless operation, overlooking that client influence is path-dependent: contributions made during training, particularly during critical learning periods (CLPs), can have disproportionate and persistent effects. Moreover, knowledge overlap and intermittent client availability make relying on all retained clients inefficient and impractical. We introduce PDF, a path-dependent client-selection framework based on the insight that retained clients are not equally valuable for unlearning. During training, PDF maintains lightweight, target-agnostic, CLP-weighted histories summarizing each client's accumulated update direction and global-model alignment. Upon an unlearning request, these histories identify and rank retained clients that are globally informative yet weakly coupled to the target, enabling retraining-based unlearning with only the top-$k$ clients over a small number of rounds. Experiments on FMNIST, CIFAR-10, and CIFAR-100 under IID and non-IID settings show that PDF achieves effective forgetting and strong retained utility with substantially fewer retained clients. In particular, PDF achieves $0\%$ forgetting accuracy with retraining-like MIA AUC on non-IID CIFAR-10, improves retained accuracy by up to $6.91$ percentage points over the strongest competing FU baseline on CIFAR-100, and reduces communication by approximately $69\%$ and computation by $57$--$61\%$ relative to Retrain.

0.33

Agreement rather than normalisation keeps depth reusable in looped language models

cs.LG cs.CL ·2026-10-22 ·Ivan Viakhirev, Kirill Borodin, Maxim Abramov, Grach Mkrtchian
hot direction
Abstract · Under review at ICLR 2027

Looped language models apply one block of layers repeatedly, and the design pays off only if depth is reusable: loop steps beyond the trained budget must remain usable at inference. We ask whether that rests on normalising the state between loop steps, as the released Ouro models do, or on agreement between the loop a model is trained with and the loop it is run with. On released weights this loop-boundary normalisation can be removed without adding parameters and with the readout path bit-identical, and the operation is an identity at one loop step, which serves as an implementation check. Removing it takes Ouro-1.4B and Ouro-2.6B to zero accuracy by the trained budget on a balanced binary task (chance 0.500). State-norm growth, the failure that prior accounts predict, is not the cause: the norm stays within a factor of 2.021 while the head commits to a single wrong token. Yet in 19M-parameter models trained from scratch on two synthetic tasks, 16 seeds per configuration, training without the normalisation is never worse than training with it and extrapolates better: the two are closest at the trained depth and separate on both sides of it (Cliff's delta up to 0.867 beyond the budget). Both results follow if agreement, and not the normalisation itself, is what keeps depth reusable. The damage sits in the readout together with the last layer of the block: in the models trained from scratch, refitting the head over the frozen modified loop recovers a third of the lost accuracy and plateaus, while refitting the last block layer with it recovers all of it. Finally, comparing looped with stacked depth at an equal inference loop count is confounded by unequal overshoot of the trained budget; at matched overshoot the advantage of looping vanishes on one task and shrinks on the other.

0.33

BeNice! Closing Looping's Inference-Memory Gap via KV-Sharing

cs.LG ·2026-10-22 ·Hayden Prairie, Jacob M. Ryan, Taylor Berg-Kirkpatrick, Daniel Y. Fu
Abstract

Looped architectures scale quality while keeping parameter count fixed by repeatedly passing activations through the same block of layers. During inference, each recurrence generates independent KV-caches, causing memory to grow linearly. To address this challenge, prior work has proposed KV-sharing (i.e., sharing KV-caches across depth), finding that, in some cases, the KV-cache of the final recurrence can be shared with the earlier recurrences without quality loss, a property we denote as KV-shareability. However, the mechanisms underlying KV-shareability remain poorly understood, with even identically trained looped architectures not consistently exhibiting it. In this paper, we study how KV-shareability can be enabled in looped models. We propose a two-stage method: in the first stage, we use stochastic depth during pretraining, which induces contractive residual dynamics that drive representations toward a common limit across recurrences. In the second stage, we propose BeNice!, a post-training method that broadcasts the final recurrence's KV-cache into the first recurrence, training it to produce compatible residuals while preserving the original model's quality. As a result, BeNice! enables KV-shareability while matching standard finetuning's unshared validation perplexity. Using BeNice!, we show that KV-sharing is complementary to alternative KV-compression methods, composing with memory-efficient sequence mixers (e.g., multi-latent attention, Kimi delta attention) and low-precision KV-cache quantization (e.g., fp8, mxfp8, nvfp4). Finally, in simulated low-memory, long-context inference, KV-sharing improves cache hit rates by up to 31.4%, increases throughput by up to 23.6%, and reduces p90 time-to-first-token by up to 83.6%.

0.33

QuaNAS: Uncertainty-Aware Reuse of Earlier Searches for Query-Efficient Neural Architecture Search

cs.LG ·2026-10-22 ·Brama Seta Putra Satria, Dwi Sunaryono, Misbakhul Munir Irfan Subakti, Irzal Ahmad Sabilla, Laras Suciningtyas
dataset
Abstract · 32 pages, 7 figures, 13 tables. Code and data: https://doi.org/10.5281/zenodo.23562488

Neural architecture search (NAS) on a new dataset often affords only tens of short training runs, so what the search knows before its first evaluation decides its efficiency. Yet most NAS methods start from scratch, guided at best by zero-cost proxies that are unreliable across tasks. QuaNAS treats a library of earlier searches in the same search space as explicit knowledge of uncertain reliability. Main effects of design decisions learned from this library and, secondarily, zero-cost proxies become priors over candidate architectures without any label of the new dataset. A Gaussian process starts at the prior optimum and re-estimates after every evaluation how far each prior is trusted. Under a protocol frozen before any confirmation data were seen, QuaNAS was evaluated once on 12 new image, time-series, text and tabular datasets. It lowered anytime regret on the short-training objective 2.7- to 3.9-fold relative to five pre-specified NAS baselines that start from scratch (Holm-adjusted p = 0.0012 each). Its final models matched (non-inferiority margin 0.5 pp) or exceeded matched-recipe reference models on nine datasets, as did the prior optimum, selected without a single evaluation. Post-hoc comparisons with established transfer methods given the same library trace the gain to the library: without it, the same Gaussian process had over twice the regret. They also show where each transfer mechanism works best: the pooled main effects of QuaNAS for weakly transferable tabular tasks, per-task surrogates for image, time-series and text tasks and for small or mismatched libraries. Finally, the main effects make the transferred knowledge inspectable.

0.33

Convex Equilibrium Potentials for Monge--Ampère: Curvature Bounds and A Posteriori Evaluation

cs.LG math.NA ·2026-10-22 ·Edison Vázquez, Yingdong Lu, Tomasz Nowicki
Abstract

Equilibrium neural networks define their hidden state as the fixed point of a contraction, so their effective depth is set at evaluation time rather than by the architecture. Using them for fully nonlinear PDEs raises two issues that explicit networks avoid: second derivatives may grow with the effective depth, and the potential is only ever evaluated approximately. We study nonnegative softplus equilibrium networks with contraction factor $q<1$, a convex class suited to the Monge--Ampère equation. A positivity cancellation in the resolvent shows that the Hessian grows at most like $(1-q)^{-2}$, rather than the $(1-q)^{-3}$ of a direct estimate, and an explicit one-neuron family shows that this exponent is sharp; third derivatives satisfy a sharp $(1-q)^{-3}$ bound. Residual-based a posteriori estimates then control the state, its input derivatives and, in two dimensions, the Hessian determinant at an approximate equilibrium. Combined with interval arithmetic on complete covers of the domain, they yield rigorous uniform bounds on the Monge--Ampère residual and, through a comparison argument, on the solution error of a fixed trained network. On two manufactured Dirichlet problems on the unit disk, all six trained models are certified with residual below $0.25$ and boundary error below $0.01$. Adaptive stopping halves float64 evaluation time at best, whereas certified adaptive evaluation remains costlier than fixed depth.

0.33

Parameter Learning in Active Inference via Variational Message Passing

cs.AI cs.LG ·2026-10-22 ·Benedikt Waldeck
Abstract · Accepted for publication in the proceedings of the 7th International Workshop on Active Inference (IWAI 2026), Springer CCIS

Active inference casts perception, learning and planning as variational inference, where the Expected Free Energy (EFE) defines the planning objective. Recent work showed that EFE minimization can be written as Variational Free Energy (VFE) minimization on a generative model augmented with epistemic priors, and that the resulting entropy-corrected objective is minimized by sum-product message passing (SP-MP) on modified factor functions. These formulations treat model uncertainty as inference over a finite set of known models. We extend them to planning during learning, where beliefs over the observation and transition parameters $θ_A$, $θ_B$ are Dirichlet priors. Closed-form Dirichlet updates are usually obtained by a structured mean-field factorization between parameters and states, but under exactly that factorization the information-gain corrections vanish. We resolve this by giving every factor its own parameter prior, frozen during a rollout. Conjugacy then makes each factor belief a product of a marginal over the categorical variables and a one-step Bayes posterior of the parameters. This reintroduces the datum-dependent parameter posteriors that mean-field factorization destroys. We show that evaluating the epistemic priors under the resulting Bethe joint recovers the information gains. Stationarity of this single closed-form objective yields SP-MP whose factor functions are augmented by the information-gain weight. Experiments on two grid POMDPs, in which only a planner that learns the model finds the safe route, show that the method plans to efficiently learn $θ_A$ and $θ_B$. It reaches the goal on every episode, whereas SP-MP planners with the same inference update plateau or take the risky route.

0.33

Extreme Compression of Per-Layer Embeddings via Cross-Layer Latent Vector Quantization

cs.LG ·2026-10-22 ·Jiacheng Liu, Yaxin Luo, Xiaohan Zhao, Peng Sun, Xinyi Shang +3 more
code/data
Abstract · Code at https://github.com/VILA-Lab/optic

Scaling embedding capacity alongside Transformer computation is an emerging trend in language models. This trend makes high compression rates increasingly useful. Hence, we study extreme compression of Per-Layer Embeddings (PLE), which follow this trend and account for 35.2% of the parameters of Gemma 4 E4B. At such rates, conventional quantizers give each token-layer vector its own code of only a few bytes. We instead develop OPTIC (Output-Path-fitted, Token-shared Indexed Codes), which stores one quantized code tuple per token and decodes it at every layer with a layer-specific affine decoder. Codes and decoders are fitted to the PLE output path rather than to the stored vectors. At about $100\times$ PLE compression of bfloat16 (BF16) E4B, OPTIC retains an aggregate 97.8% of the uncompressed model's performance across seven benchmarks, whereas product quantization, the strongest of six conventional baselines, retains 66.3%. At $594\times$ and $1,002\times$ PLE compression of BF16 E4B, OPTIC retains 90.5% and 84.1%. On the official E4B release with 4-bit linear weights and 16-bit activations (W4A16), OPTIC retains 96.9% at about $100\times$ PLE compression. During inference at about $100\times$ PLE compression, BF16 E4B uses 29.7% less GPU memory than with uncompressed PLE at the same key-value cache size, while running at most 0.40% slower.

0.33

Who Is the Influencer? Consensus Dynamics in Multi-Agent Systems

cs.AI ·2026-10-22 ·Gouki Minegishi, Jingyuan Feng, Hiroki Furuta, Takeshi Kojima, Yusuke Iwasawa +1 more
Abstract

Multi-agent LLM systems are increasingly used in practice, but their behavior is complex and hard to understand. We design a simple task to ask which model has the most influence on the final consensus when several models are mixed in one system. Experimentally, across seven model families, we find that the opinions of the other models are absorbed into the opinion of the stronger model, the one with more parameters. This happens even though the task requires no pretraining knowledge and has no correct answer. To understand this phenomenon, we interpret the agents as weighted particles moving on a simplex. Larger models carry a larger weight and therefore occupy a deeper well in the energy landscape. A large weight means strong in-context learning: a model with a large weight amplifies the majority of its context. Larger models are generally better at in-context learning, and this is what makes the larger model the influencer. We hope that this work helps to interpret multi-agent systems, whose behavior is becoming too complex for humans to understand.

0.33

TACS-RF: Tool-driven Agentic Circuit Synthesis of RF and mmWave Modules

cs.AI ·2026-10-22 ·Tristan Liang, Aravind Nagulu
published
Abstract · Accepted for presentation at the 2027 IEEE Radio and Wireless Week (RWW), Silicon Monolithic Integrated Circuits in RF Systems (SiRF)

We present the framework of TACS-RF - a Tool-driven Agentic Circuit Synthesis of RF and mmWave modules. In this framework, a large-language-model (LLM) coding agent (Claude Opus 4.8) directly drives industry-standard electronic-design-automation (EDA) tools such as Cadence Virtuoso/Spectre, Calibre, Cadence EMX, and Ansys HFSS to design RF and mmWave modules. We validate the framework through the agentic design of a 240-251 GHz sub-THz transmitter in a 90 nm SiGe BiCMOS process. The agent semi-autonomously designed the sub-blocks including a 60 GHz push-push Colpitts VCO with second-harmonic extraction, multistage differential cascode power drivers, a push-push frequency doubler, a differential bowtie antenna, and all the interstage matching networks. The transmitter chain with parasitic-extracted devices and EM-verified passives delivers +5.1 to +6.3 dBm across an 11 GHz varactor-only tuning range, phase noise of -80 to -88 dBc/Hz at 1 MHz offset while consuming a DC power of 361 mW. We describe the agent-driven design flow and the resulting division of labor between human and agent, taking a step toward a "text-to-ASIC" future for designing RF/mmWave hardware.

0.32

Efficient and Optimal Learning of Discrete Distributions with Person-Level Privacy

cs.LG cs.CRcs.DS ·2026-10-22 ·Gautam Kamath, Mahbod Majid, Ankur Moitra, Argyris Mouzakis, Jonathan Ullman
publishedtheory
Abstract · Full version of the SODA 2027 paper

We study learning a discrete distribution under person-level differential privacy in the batched setting. Given $n$ users, each contributing $m$ i.i.d. samples from an unknown distribution $μ$ over $[k]$, we seek an estimate of $μ$ with total-variation error $α$. We give polynomial-time algorithms under both pure and approximate differential privacy. Under pure $(\varepsilon,0)$-DP, we achieve $n = \widetilde O(k/(α^2 m)+k/(α\sqrt{m}\varepsilon)+k/\varepsilon)$, the first polynomial-time pure-DP guarantee matching the information-theoretic optimum up to logarithmic factors. Under approximate $(\varepsilon,δ)$-DP, we achieve $n = \widetilde O(k/(α^2 m)+k/(α\sqrt{m}\varepsilon)+k+\log(1/δ)/\varepsilon)$. At $δ=2^{-Θ(k)}$, this is $\widetilde O(k)$ for fixed $α,\varepsilon,m$ and failure probability. Our algorithms use a new degree-6 sum-of-squares estimator for learning from untrusted batches. Its key ingredient is a trace-bounded diagonal majorizer for the difference between the empirical covariance of the repaired batch averages and the multinomial covariance predicted by their mean. We prove that clean data admit this certificate through semidefinite-programming duality and a new application of the stronger symmetric form of Grothendieck's inequality due to Friedland and Lim. The certificate preserves the mean-dependent variance of multinomial batch averages, yielding the near-optimal robust rate $O(η\sqrt{\log(1/η)/m})$ and providing the score required by the privacy reduction. We also prove a statistical query lower bound ruling out a uniformly smaller order of error under fixed polynomial SQ resource bounds.

0.32

Speculative Diffusion is a Tale of Two Tradeoffs

cs.LG ·2026-10-22 ·Aaron Zorel, Guy Van den Broeck, Aditya Grover, Daniel Israel
Abstract

Speculative decoding is a technique that accelerates the inference speed of autoregressive large language models (LLMs) by sampling from a faster, lower quality draft model. Traditionally, a fundamental characteristic of speculative decoding is that it is lossless with respect to the verifier distribution, guaranteeing speed without sacrificing quality. This differs from recently developed diffusion LLMs (dLLMs), which are notably able to achieve substantial speedups by degrading the quality of their outputs with less diffusion effort. Because diffusion models are able to sample faster than autoregressive models by generating tokens in parallel, they are considered strong candidates for draft models in speculative diffusion. While it is widely understood that diffusion models can sacrifice quality for speed, speculative decoding itself can also be generalized to generate lossy samples faster using more lenient acceptance criteria with respect to the verifier distribution. Presented with the two parameters of diffusion effort and speculative decoding leniency, this paper explores the full achievable Pareto frontier via joint variations of these parameters. Our experiments show that jointly varying diffusion effort and acceptance leniency can extend the speed-quality Pareto frontier over single step diffusion drafting.

0.32

Solving Flood-It Puzzles: A Comparison of Tailored A* Search with Declarative Solver Approaches

cs.AI ·2026-10-22 ·Philipp von Manteuffel, Benjamin Blankertz
Abstract · 22 pages main text, 35 pages in total; 13 figures plus 12 figures in the appendix

We present an optimized approach for finding minimal move sequences to solve Flood-It puzzles using A* search and compare it with approaches using SAT, Integer Programming, and Constraint Programming solvers. The performance of A* search is considerably improved by using an appropriate heuristic and sophisticated equivalences of game states to limit the search space. Each aspect is systematically evaluated, providing insights that generalize to other problems. By comparing the best variant of each approach, the A* search is the fastest in finding a minimal solution in about 75 % of the Flood-It instances tested. This result is in line with our expectation since A* specifically looks for a minimum solution. In contrast, the SAT-based approach is designed to solve the decision problem of whether a solution with a given length exists. An iterative search is required to determine a minimum solution length using a SAT-based approach. However, with increasing complexity, quantified by the minimum solution length, the SAT-based approach outperforms A*. We also compared the different approaches to the decision problem, where A* again performed best for the largest number of puzzles. Here, again, SAT supersedes A* for higher complexities, albeit with a lower crossing point.

0.32

CRANE: Post-Hoc Counterfactual Retrieval for Multimodal Classifiers

cs.LG ·2026-10-22 ·Franco Rugolon, Ioanna Miliou, Panagiotis Papapetrou
published
Abstract · 7 pages + references, 3 figures, accepted to DSAA 2026

Counterfactual explanation methods have been introduced for unimodal settings, with dedicated frameworks for tabular, timeseries, text, and image data, yet no post-hoc method has been designed for classifiers that learn from multimodal inputs. Existing explainability methods for multimodal classifiers apply modality-specific attribution independently to each input, which fails to yield coherent explanations and does not transfer to arbitrary pre-trained multimodal architectures. To address these shortcomings, we present CRANE (Counterfactual Retrieval via Agnostic Nearest-neighbour Explanations), the first post-hoc, model-agnostic framework for nearest-neighbour multimodal counterfactual generation that retrieves real training examples from the target class as candidate explanations, ensuring plausibility by construction. Across four heterogeneous datasets and three fusion strategies, we show that Multimodal Consensus Retrieval (MC-R), which intersects per-modality neighbour sets to identify instances jointly close to the query in every modality, emerges as the strongest multimodal strategy, achieving the best or joint-best outcome on three of four datasets while grounding every explanation in the joint training distribution. Unimodal strategies remain competitive when a single modality dominates the decision boundary.

0.32

Model Organisms of Collusion: Corrective Transfer in Untrusted Monitors

cs.AI ·2026-10-22 ·Juan-Pablo Rivera, Ari Spiesberger, Juan J Vazquez, Arun Jose, Axel Ahlqvist +4 more
Abstract · 33 pages, 6 figures, 15 tables

AI control protocols aim to get useful work from models that may be misaligned by monitoring their behaviour. One approach uses a second copy of the same model as the monitor. This avoids relying on a weaker reviewer the model could outsmart, but risks collusion: the model, now monitoring itself, may cover up its own misbehaviour. One remedy is to fine-tune the monitor on whatever examples of misbehaviour the control team has caught. We ask whether this training also removes collusion that the control team has not yet found. We fine-tune Llama-3.3-70B and Qwen2.5-72B to install five policy-monitor behaviour pairs, correct the monitor on 300 honest reviews of one caught collusion, and measure collusion on the other four in unseen prompt formats. Averaged over the three seeds of each family, correction reduces held-out collusion from 91% to 14-20% under three of the four methods we compare (individual seeds 10-32%), while a control fine-tuned on the same number of unrelated examples leaves 56-93%. Family-average accuracy on control items changes by at most five points and capability is preserved, apart from a drop in mathematical accuracy on one Llama seed. Under the tested training configuration, methods with similar caught-task accuracy differ in held-out transfer, from 58% to 83% average reduction, so the caught behaviour alone does not reveal how far a correction reaches. Training on a known failure removes it almost entirely (1-2% remaining), yet 14-20% of unknown failures persist, so near-perfect correction of the caught behaviour overstates how much collusion has been removed.

0.32

Federated Differentially Private Optimization via Low-Rank Preconditioning

cs.LG cs.CR ·2026-10-22 ·Antti Koskela, Tejas Kulkarni, Elias Raninen
hot direction
Abstract

We consider differentially private (DP) optimization in the federated setting, where a DP estimate of the feature covariance matrix serves as a Hessian approximation for generalized linear models. Existing central DP methods for covariance estimation have communication cost that grows quadratically in the data dimension, while existing baselines with linear communication cost are only locally private and degrade as the number of clients increases. As a remedy, we propose a federated DP-PCA framework that achieves linear communication cost while providing central DP guarantees under secure aggregation. It consists of a DP rank selection step, a DP power method, and a noisy release of the covariance projected onto the selected low-dimensional subspace. We give error bounds for this covariance estimate and for the resulting low-communication preconditioned DP gradient descent. Empirically, our proposed DP optimization method matches a centralized baseline with quadratic communication cost at a fraction of the communication.

0.31

Energy and Latency First: How Deployment Constraints Reverse Time-Series Model Selection

cs.LG ·2026-10-22 ·Shruti Bothe, Illyyne Saffar, Ayush K. Varshney
publishedbenchmark
Abstract · 13 pages, 1 figure

AI models, and time-series models in particular are typically ranked by accuracy and deployed under constraints absent from that ranking. In this paper, we study a real setting where these constraints are explicit and measurable: handover pre- diction in cellular networks, where inference runs on the User Equipment once per 40 ms measurement period and must complete before the network acts. The timing bound is hard: a late prediction is superseded by the measurement it was predicting, while energy is a cumulative budget spent for the life of the device. Neither appears in the objective these models are selected on. We evaluate eight architectures under a protocol that screens for deadline feasibility at the tail rather than the mean, then ranks the feasible models on measured inference energy, count- ing the feature computation that precedes the forward pass and is usually omitted. Gradient-boosted trees on handcrafted temporal features are the most accurate model (F1=0.974) and among the cheapest to run, while the neural architecture an accuracy-only benchmark would rank first (CNN-1D, F1=0.914) costs over 3 times more energy per inference under a realistic duty-cycled deployment pattern. We give three conditions under which this outcome should be expected, so the finding applies beyond telecommunications. Accuracy-only benchmarking does not merely under-report deployment cost. It can rank first a model that cannot be deployed.

0.31

Rethinking Sequence Compression with Test-time Regression

cs.LG ·2026-10-22 ·Sihan Yang, Jie Huang, Hanxuan Li, Yifan Guo, Weigao Sun +2 more
code/data
Abstract · Technical report (19 pages, 4 figures, project page: https://spherelab.ai/ttr-compressor)

Sequence compression is a fundamental operation and a critical information bottleneck in efficient language and multimodal models, and yet prevailing approaches based on pooling or learned projections remain underperforming. We revisit sequence compression through the lens of test-time regression: input tokens define a regression problem whose fitted coefficients form the compressed representation. Importantly, the regression problem is learned end-to-end via backpropagation, while its coefficients are computed efficiently in closed form during the forward pass. This leads to TTR-Compressor, a general alternative to conventional pooling and projection-based compression. We evaluate TTR-Compressor as the compressor for block-wise sparse attention and the visual token merger for multimodal large language models. On block-wise sparse attention, TTR-Compressor improves RULER performance over attention-based compressors by a large margin. For visual token merging, it consistently outperforms the standard MLP merger and matches its performance with 55.6% fewer visual tokens. These results establish test-time regression as an effective primitive for sequence compression in efficient language and multimodal models.

0.31

First hardware demonstration of a continuous-state probabilistic computer with programmable energy-defined sampling

cs.NE cs.ET ·2026-10-22 ·Chao Fang, Hsuan Yao Tu, Yihan He, Xiao Gong, Chin-Fu Nien +1 more
Abstract · 22 pages,5 figures

Probabilistic computing has proved useful for combinatorial optimization, but conventional implementations rely on binary stochastic units and sigmoid-based updates. Extending these systems to continuous variables requires binary encoding. Here we introduce a continuous probabilistic computer (CPC) whose continuous probabilistic bits sample bounded states from conditional distributions defined by local energies. Separating random-source generation from conditional sampling provides a route to using physical random sources with different native distributions. A field-programmable gate array (FPGA) prototype updates finite-precision states using digital randomness. The same framework supports product constraints, memory recovery and solution-space exploration. For the tested 7-bit factorization case, fixed-temperature FPGA CPC reduces mean core solving time approximately 33,000-fold relative to the evaluated quadratic unconstrained binary optimization software implementation. The FPGA recovers stored MNIST patterns without pretrained networks, achieving final success rates of 100%. Software CPC reaches 90% reference coverage of the feasible area in approximately 18% of the time required by restarted Adam.

0.31

Not Every Agent Needs the Strongest Model: Unified Prompt-Model Optimization for Heterogeneous LLM-based Multi-Agent Systems

cs.MA ·2026-10-22 ·Mingxiu Cai, Haoyang Xian, Xinyu Ma, Yutong Wang, Zhenxi Song +2 more
code/data
Abstract

Large language model (LLM)-based multi-agent systems (MAS) rely on role-specific prompts and execution models to solve complex tasks collaboratively. However, prompt optimization typically keeps model assignments fixed, leaving differences in role requirements and their associated inference costs outside the search. We present UniMASO, a Unified framework for heterogeneous Multi-Agent System Optimization through the joint adaptation of prompts and model assignments. UniMASO evaluates candidate updates through local, downstream, and global feedback, using cost-aware Pareto search to retain useful quality-cost trade-offs. Topology-aware co-evolution then coordinates these updates according to agent roles and dependencies. Cross-level trajectory memory further supplies historical successes and failures to guide subsequent optimization. Experiments on five benchmarks and five execution topologies show that UniMASO achieves competitive task performance at substantially lower deployment cost than the strongest homogeneous configurations. Under centralized execution, it improves the average score from 69.79 to 70.40 while reducing token-weighted deployment cost by 85.2% relative to DeepSeek-V4-Pro with MASPO. These results support joint prompt-model adaptation as an effective approach to efficient multi-agent collaboration. The code is available at https://github.com/Achilles008/UniMASO_main.

0.31

What Lies Beyond Perception: A Partition Theory of the Limits of AI Evaluation and Oversight

cs.AI cs.CYcs.LG ·2026-10-22 ·Ahmed M. Mahran
benchmarktheory
Abstract · 36 pages (23 main text, then references and appendices), 6 figures, 2 tables

Replacing the judge can move a benchmark score, and models can be trained to hide capabilities. Both put the evaluator in question: what can it certify? We develop a partition theory of evaluation from one principle: judgement is bounded by perception. An agent's discrimination is the partition of its inputs into classes it cannot tell apart. Discrimination is necessary for capability but not sufficient, so bounds on it bear directly on whether evaluation can show that a model lacks an ability. We prove a one-sided bound. An examinee whose partition refines its evaluator's has the discrimination to pass every test the evaluator can grade, provided each input has an acceptable reply. We define the echo test, which accepts any reply the evaluator perceives as identical to the input, and show that it is complete for the tests the evaluator can grade. Passing therefore certifies at most a floor on discrimination, never a ceiling, and only a sharper evaluator can raise it. Used to detect concealment, echo-test failure points the wrong way: under exhaustive probing it flags every honestly weaker examinee and no concealer that presents as the evaluator's peer. Exact imitation of a weaker model is undetectable by any behavioural evaluator. We extend the bound to ensembles of monitors and to channels a model controls, explain published findings on judge disagreement and sandbagging, and derive testable predictions. A safety case that argues from behavioural evaluation alone that a model lacks an ability must assume the model's cooperation.

0.31

SCOPE: Compiling Natural-Language Policies into Neurosymbolic Predicate Programs

cs.AI cs.CL ·2026-10-22 ·Arnon Mazza, Elad Levi
benchmark
Abstract

Policy enforcement for LLM agents has shifted from single-turn content moderation to compound, application-specific policies evaluated over multi-turn transcripts and tool-call trajectories. Fine-tuned guardrail models are efficient but degrade on compound policies, and while LLM-as-a-judge approaches achieve higher accuracy, they incur high inference cost and tend to produce unstable predictions. Neurosymbolic methods offer a natural alternative, decomposing the policy into narrow neural decisions composed under explicit logic. However, existing symbolic layers lack the required expressiveness and are optimized against fixed data rather than the policy itself. To address this, we introduce SCOPE, which compiles a policy into an executable program of neural predicates executed by small models, deterministic extractors, and code that composes their outputs. The program is optimized jointly with its training data, using adversarial generation of counterexamples that target where the program deviates from the policy. We evaluate SCOPE on a benchmark of real-world policy tasks against prompt optimization, task decomposition, and workflow generation methods, achieving the highest average accuracy across tasks and models. Programs optimized with the weakest executor transfer to other models without further optimization. This allows them to be deployed in production on inexpensive models, where they reach accuracy comparable to frontier reasoning models. Unlike these monolithic models, SCOPE produces substantially more stable predictions, and each verdict is interpretable through the predicate outputs that produced it. Additionally, we validate the key components of our method through an ablation study. SCOPE thus enables reliable, interpretable, and efficient policy enforcement at production scale.

0.31

TESSA: A Governance-Oriented Custom GPT for SME Guidance under the EU AI Act: Design, Implementation, and Evaluation

cs.AI ·2026-10-22 ·Chalisa Veesommai Sillberga, Kai-Kristian Kemella, Pekka Sillberga, Katri Harjuveteläinenb, Mika Saaria +1 more
publishedbenchmark
Abstract · 18 pages and 2 figures. Accepted for presentation at the 17th International Conference on Software Business (ICSOB 2026)

The increasing adoption of generative AI presents new opportunities while raising AI governance and regulatory challenges, particularly for small and medium-sized enterprises (SMEs) navigating the European Union Artificial Intelligence Act (EU AI Act). To support AI governance education, this study developed TESSA, a governance-oriented educational artifact within the OpenAI Custom GPT environment. Following a design science research approach, TESSA was evaluated against baseline ChatGPT using 100 benchmark scenarios through governance alignment assessment, human expert review, and consistency testing. Preliminary LLM-assisted assessment showed governance alignment rates of 90% for TESSA and 83% for baseline ChatGPT, while independent validation by an AI regulation law practitioner yielded 71% and 70%, respectively. Human expert review indicated that both systems provided generally strong governance guidance, while TESSA maintained high consistency across repeated executions. The findings suggest that governance-oriented grounding can support structured, practical, and stable AI-assisted compliance guidance for SMEs. The study contributes (i) a governance-oriented Custom GPT artifact for AI governance education, (ii) a human-validated regulatory update mechanism, and (iii) a multidimensional evaluation framework for governance-oriented AI artifact.

0.31

Co-Recursive Semantic Self-Extension: A Mathematical Foundation for Open-Ended Intelligence

cs.AI ·2026-10-22 ·Simon Hegelich, Bardh Prenkaj, Habiba Sarhan, Morteza Shahrezaye, Kolja Hegelich +2 more
Abstract

Artificial intelligence has advanced through scale, data, and computation, yet remains bounded by its representational framework. We ask what makes intelligence open-ended. Every inferential system selects meaningful distinctions, limiting what it can learn. Within a fixed purpose, performance reaches a ceiling set by its semantic domain. Teleological succession, in which purposes arise from earlier ones, faces a similar limit: Boolean-closed purpose-transformations exhaust their reachable semantic territory. We shift from purposes to inferential systems. Drawing on Kant and Gödel, we model cognitive states as partial evaluations over a formal language. Unresolved propositions may be undecided or undecidable. Reasoning can resolve the former but not the latter, leaving a frontier for extension. Non-trivial reductions of undecidable propositions expose undecidable intermediate premises whose reflective reification supports multiple decisive extensions. Finite trace-guided generation remains fallible, requiring meta-level selection and polycontextural exploration. Using Kolmogorov complexity, we show that no architecture with a fixed state description, transition mechanism, and initial condition can generate histories incorporating irreducible algorithmic information. Because physical systems are fixed in this sense, we examine collectives. Coupling finitely many finite agents does not remove the bound. Extension requires persistent external memory preserving semantic content and historical coordinates. Collective generative capacity depends on balancing overlap for transfer with difference for generation. We propose semantic self-extension as a design principle: intelligence should not merely resolve ambiguity within existing meanings, but transform unresolved structures into new spaces of meaning, generating new distinctions, questions, representations, and purposes.

0.31

Disentangling Robustness and Sycophancy in Multilingual LLMs

cs.AI cs.CL ·2026-10-22 ·Rhitabrat Pokharel, Ameeta Agrawal
hot direction
Abstract

When a multilingual LLM agrees with a user's false premise, is it being sycophantic, or does it simply not know better in that language? Current research does not distinguish these cases: they record only whether the model agreed, not why. We introduce the SC-matrix, that crosses a model's competence (whether it answers correctly under no pressure) with its sycophancy (whether it maintains that answer under pressure). Our results show that (1) identical aggregate sycophancy rates can mask different failure compositions, (2) high competence in a language does not always protect against sycophancy, and (3) reasoning-oriented mitigations outperform directive-based strategies. We find that multilingual sycophancy is not a single phenomenon but a composite of different failures that require distinct diagnosis.

0.30

RobustNCL: Robustifying Neuromorphic Continual Learning under Transient Faults with Lightweight Detect-and-Cure Strategy

cs.NE cs.AIcs.LG ·2026-10-22 ·Rachmad Vidya Wicaksana Putra, Shuochen Zhou, Muhammad Shafique
hot direction
Abstract · 12 pages, 6 figures, 3 tables

In neural networks, continual learning (CL) is a prominent approach for enabling deployed systems to adapt to dynamic operational environments by effectively learning from new unseen data at runtime. Recently, the neuromorphic continual learning (NCL) concept has emerged as an energy-efficient alternative to CL by leveraging sparse event-based operations. However, NCL performance can be undermined by transient faults in the hardware platforms, which may come from high-energy particle strikes at any time. Currently, NCL reliability under transient faults has not been studied, and hence the corresponding fault-tolerance techniques remain unexplored. To address this, we propose a novel RobustNCL framework to improve NCL reliability in the presence of transient faults in the DRAM-based off-chip memory of hardware platforms through a lightweight detect-and-cure strategy. Specifically, it first monitors the presence of faults during weight data access from off-chip memory, and then cures the accessed weights if anomalous signatures are detected. It provides three curing variants: replacement with zero, min/max value, and most-probable value from the original/clean weight distribution. To support this, the weight curing values are stored in the radiation-hardened memory. Experimental results show that, in the presence of high fault rates (e.g., 5x10^{-4} for memory replay), RobustNCL maintains high accuracy close to the accuracy of fault-free models (i.e., within -0.5% accuracy for the best RobustNCL variant) and significantly improves accuracy from the baseline without fault mitigation across different NCL methods and datasets (i.e., Split MNIST, Split Fashion-MNIST, Split CIFAR-10, and Split CIFAR-100), while incurring negligible latency and energy overheads. These results highlight that our RobustNCL framework is a promising approach for enabling reliable and energy-efficient NCL systems.

0.30

Forecasting Associations, Not Interventions: A Preregistered Causal Audit of Covariate-Aware Time Series Foundation Models

cs.LG ·2026-10-22 ·Zong-Han Bai
benchmark
Abstract · 21 pages, 11 figures, 12 tables

Covariate-aware time series foundation models (TSFMs) accept a future path for a control variable and return a forecast, inviting what-if queries. The query is interventional; the models are trained on observational data. We determine whether their answers are the conditional slope $\partial E[Y|X]$ or the interventional effect $\partial E[Y|do(X)]$, at three levels. Theory: a Bayes-optimal observational predictor answers interventional queries associationally unless the adjustment set is available and used; for in-context least squares we derive the resulting confounding gap -- a model's effect estimate under an observational minus under a randomized context -- in closed form, a parameter-free reference line. Behaviour: on a preregistered grid with exact counterfactual ground truth, Chronos-2, TiRex-2, TabPFN-TS and TimesFM-2.5 all track this line ($R^2$ 0.90-1.00) and respond more confounded than regression itself (slopes 1.3-1.8), on linear and nonlinear structural models. Pairs of models with identical observational laws prove this is not an identifiability limit: when the effect is identifiable, least squares recovers 100% of the true effect difference, Chronos-2 75%, TiRex-2 34%. Representation: linear probes decode the data-generating regime (94-96%) and the true effect ($R^2$ 0.7-0.9) from embeddings that the forecasts ignore. On two physical systems with randomized actuators, pretrained models under-respond by 2-5x and ignore in-window confounding. Minutes of interventional post-training on 3,000 synthetic series cut the gap by up to 83% and transfer outside the training prior. Its forecasting cost is a replay-distribution effect: replaying the model's own pretraining prior cuts it from +47% to +10% MASE, and with low-rank adaptation yields a recipe meeting the preregistered criterion over three seeds (52-61% gap reduction at +1-3% cost) while improving the DoTime benchmark.

0.30

What Makes a Good Unmasking Policy? A Controlled Study of Learned Unmasking in Diffusion Language Models

cs.AI ·2026-10-22 ·Jinya Sakurai, Xun Xu, Jaehong Yoon
Abstract · Preprint.32 pages, 11 figures

Masked diffusion language models generate tokens in parallel, but their efficiency depends on which predictions are committed. Learned unmasking policies improve the speed--accuracy trade-off, yet differ in supervision, inputs, and capacity, obscuring which choices matter. We view supervised policies as distilling a decoding procedure through a teacher and examine independently what the teacher labels as committable, on which states the policy imitates it, what the policy observes, and how much the number of function evaluations (NFE) must decrease to offset policy overhead. We find that labeling predictions matching the final output is overly permissive at low NFE, and hidden states are weak alone but can complement confidence. About a 10% NFE reduction suffices to offset policy overhead for policies up to 7.5M parameters across tested settings. We therefore train a policy on trajectory-contraction targets, which accept only commitments staying on a reference trajectory, using confidence and hidden states. On LLaDA-8B-Instruct, the resulting 856K-parameter policy achieves the best accuracy--NFE frontier and highest throughput on GSM8K, is competitive at moderate targets on MATH-500 and HumanEval, and largely transfers to Dream-7B-Instruct.

0.30

Diagnosing the Explicit-to-Applied Theory of Mind Gap in LLMs

cs.AI ·2026-10-22 ·Prajvi Saxena, Sabine Janzen, Wolfgang Maass
benchmark
Abstract

As large language models are deployed in high-stakes roles such as therapy, education, and personal assistance, their social cognition must be assessed reliably. Current Theory of Mind (ToM) benchmarks report aggregate scores that conflate two distinct competencies: explicit ToM (recognizing what someone believes) and applied ToM (using that belief to predict behavior). Building on this distinction, we investigate the gap and its recoverability using four structurally diverse benchmarks. Aggregate reporting obscures uneven competence: on SimpleToM, Llama-3.1-8B scores 51.5 percent overall while recognizing mental states at 99.4 percent and predicting behavior at only 28.9 percent. To assess recovery, we compare structured prompting, matched in-context learning, and our proposed Belief-State Test-Time Training (BSTTT), which adapts lightweight parameters on answered support questions through action reconstruction. Across Qwen2.5-7B, Llama-3.1-8B, and GPT-oss-20B, recovery depends on the model, task, and intervention. On SimpleToM, reminders combined with BSTTT improve Llama's behavior prediction by 40.8 percentage points; reminders account for most of the overall improvement, with BSTTT adding 1.3 to 5.5 points beyond reminders across models. Contextual examples often outperform weight updates, while BSTTT remains stronger in specific model-benchmark settings. Temporal belief tracking remains difficult despite partial recovery. These findings show that aggregate ToM scores can overstate applied competence while hiding performance recoverable through targeted assistance. We recommend decomposed reporting of explicit and applied accuracy alongside intervention comparisons that establish which failures are recoverable and under what conditions.

0.29

SetCommit: Coordinated Token Commitments for Few-Step Diffusion Language Model Decoding

cs.AI ·2026-10-22 ·Yuannuo Feng, Yubing Ye, Wenshuai Yao, Yizhe Chen, Ngai Wong +2 more
code/data
Abstract

Diffusion language models generate multiple token positions in parallel, but few-step decoding makes parallel commitment difficult: tokens that appear reliable individually may be jointly inconsistent, while premature commitments can prevent later context-dependent refinement. In this paper, we introduce SetCommit, a training-free method for coordinated parallel token commitment. Instead of selecting tokens independently by confidence, SetCommit selects a commitment set that balances token reliability against local joint commitment risk, preserving conditioning opportunities for subsequent denoising steps. It further performs budget-neutral local exchanges, replacing vulnerable earlier commitments when newly revealed context supports more reliable alternatives. SetCommit requires no retraining or architectural modification and is compatible with existing diffusion decoding strategies. Experiments with LLaDA-8B and Dream-7B on GSM8K, MATH-500, MBPP, and HumanEval show accuracy/pass@1 gains of up to 7.4 percentage points over native decoding at 32 model evaluations. At common quality targets, geometric-mean incremental speedups range from $1.44$ to $1.74\times$. Code is available at https://github.com/Kissmetothemoon/SetCommit.

0.29

Neural Spectral Capacity with Residual Context for Heterogeneous Resource Allocation within Transformers

cs.LG cs.AI ·2026-10-22 ·Chenyu Zhu, Ruoyu Zhao, Zhichao Lu
code/data
Abstract

Predicting relative trained performance from architectural specifications can guide design when training costs limit experimentation. Neural Spectral Capacity (NSC) provides such a heuristic, but scores projections independently and assigns identical scores to compatible reorderings of the same layer configurations. We introduce Residual-Aware Neural Spectral Capacity (ResNSC) for Transformers, using a specification-derived statistical reference for accumulated residual updates to determine the input and residual-state statistics in each projection's capacity calculation, while retaining additive evaluation without data or instantiated weights. When residual widths vary, earlier choices affect later scores, so we augment NSC's allocator (NSC-DP) with residual history beyond its layer-and-budget state. ResNSC-DP jointly allocates residual and feed-forward widths under a resource budget, with certified bounds on the best numerically evaluated ResNSC score within a specified finite candidate set. At the 121M and 374M pretraining scales, selected architectures achieve lower mean perplexity than uniform, variable-width, and tapered allocations on FineWeb-Edu, WikiText-2, and LAMBADA at comparable parameter budgets. In inherited-weight LLaMA-7B pruning using the LoNAS supernet, ResNSC-DP improves eight-task average accuracy over NSC by 3.28--5.46 percentage points across eight matched parameter budgets. Paired reversals preserve the projection shapes and NSC scores yet worsen performance in both settings, exposing performance differences hidden by position-independent scoring. The code for this work is available at https://github.com/Optima-CityU/neural-spectral-capacity.

0.29

Towards Adversarially Robust Multigraph Neural Networks for Financial Crime Analysis

cs.LG ·2026-10-22 ·Dennis Heijmans, H. Çağrı Bilgi, Rui Wang, Kubilay Atasu
publishedcode/data
Abstract · 20 pages, 10 figures, 3 tables. Accepted at the Fifth Learning on Graphs Conference (LoG 2026). Code: https://github.com/denheijmans/robust_…

Financial transaction networks are naturally represented as multigraphs, where repeated transactions between the same pair of accounts are modeled as distinct edges. Recently proposed multigraph neural network architectures leverage this structure explicitly to achieve a strong performance on anti-money laundering (AML) tasks. However, their robustness to adversaries that manipulate transaction structures to evade detection remains largely unexplored. We present the first scalable framework for structural attacks and adversarial training of multigraph neural networks using node-edge incidence matrices, adapting projected gradient descent and projected randomized block coordinate descent methods to optimize structural perturbations over such matrices. We further introduce a cycle-based regularizer that penalizes changes in the global counts of directed 2- and 3-cycles, providing a structural measure of attack plausibility. Experiments on synthetic pattern-detection and realistic AML datasets show that our learned attacks substantially outperform non-learnable perturbation baselines in reducing detection performance (PR-AUC), while adversarial training consistently improves robustness against these attacks. These results demonstrate that multigraph neural networks are vulnerable to learned structural attacks and that adversarial training provides an effective defense. Our code is available at https://github.com/denheijmans/robust_multignn.

0.29

Inherited, Not Invented: Tracing the Provenance of Wait in Reasoning Models

cs.LG cs.CL ·2026-10-22 ·Zhizhuo Kou, Sizai Hou, Long Chen, Jian Yang, Junyu Luo +4 more
Abstract · Under review

Reinforcement learning (RL) is credited with teaching reasoning models to write {wait} before reconsidering a step, yet large RL accuracy gains do not require the word. This view can confuse what RL discovers with what a model inherits from its training data. We trace the word through open models' training histories, reading it with label-free indices of frequency, position, form and usage and an injected-error test, all computed from sampled outputs. We find that zero-RL almost never coins the marker: on bases that lack the word, RL raises MATH-500 accuracy substantially while {wait} stays near absent. It arrives with training data: mid-training in OLMo-3, and fine-tuning on reasoning traces installs it in most models whose base lacks it. Where the base already uses it, both released RL-Zero recipes (OLMo-3, MiMo) multiply it, whereas KL-free GRPO from the same base does not; OLMo-3's recipe wears down its form and makes it a more sensitive but more liberal alarm; RL after supervised fine-tuning barely changes it, and our GRPO recipe erases a marker seeded by fine-tuning. Finally, we show that marker frequency alone does not indicate RL-induced reasoning and caution against LLM-as-a-judge revision metrics, as automated judges can reverse human-labelled trends.

0.29

Random Projections for Spectral Algorithms: Optimal Rates in Sobolev Norms

cs.LG stat.ML ·2026-10-22 ·Caixing Wang, Yue Wang
Abstract

Random projections (RP) offer an effective method for reducing computational and storage overhead while preserving the geometric structure of data. However, existing studies have primarily focused on the optimal generalization performance of specific kernel-regularized algorithms with RP in the well-specified setting and under restrictive conditions. In this paper, we provide a comprehensive and improved analysis of the generalization performance of RP-based spectral algorithms under general conditions \change{without sacrificing statistical efficiency}. By leveraging the embedding property of the reproducing kernel Hilbert space (RKHS) and a refined analysis of the operator similarity, we establish optimal learning rates in Sobolev norms that match the minimax lower bounds up to logarithmic factors. For two common implementations of RP, randomized sketches and Nyström sub-sampling (uniform or leverage-based), we show that the projection dimension required for optimality is proportional to the average or maximal effective dimension, yielding a significant reduction in computational cost while maintaining the statistical efficiency. Our results do not rely on the uniform boundedness assumption on the regression function and are valid across a wide range of source conditions, i.e., $s> α_0-1/β$, where $s,β$, and $α_0$ denote the smoothness index, capacity index, and the embedding index, respectively. In the benign case when $α_0=1/β$, the optimality holds for all $s\in (0,2τ]$ with $τ$ and $γ$ denoting the qualification index and norm index. Experimental results demonstrate both the theoretical validity and the practical effectiveness of the proposed method.

0.29

Graph-Recurrent Learning of Emergent Collective Dynamics

cs.LG cond-mat.str-elcond-mat.stat-mech ·2026-10-22 ·Ali Rayat, Acacia Lee, Gia-Wei Chern
unusual
Abstract · 13 pages, 8 figures

Collective ordering in correlated electron systems emerges from microscopic dynamics, yet the evolution of observable textures need not admit a closed, memoryless description. We introduce a graph-recurrent neural network (GRNN) framework that learns such emergent dynamics directly from trajectories without specifying microscopic equations of motion. The framework integrates gated recurrent memory with graph message passing: local hidden states encode information beyond the instantaneous observables, while spatial coupling allows this information to propagate between sites. The observable and hidden fields evolve jointly through a shared local rule, providing a scalable representation of dynamics with spatial interactions and temporal memory. We demonstrate this framework on the emergent dynamics of charge-density-wave order in the semiclassical Holstein model, successfully reproducing temporal and spatial correlations and dynamical scaling during domain coarsening. By learning finite-time evolution directly, the GRNN offers a route to efficient, data-driven descriptions of collective dynamics with unresolved microscopic degrees of freedom.

0.29

Tracing Subliminal Learning Through Pretraining

cs.LG ·2026-10-22 ·Angela Tang, Vishesh Anand, Gonçalo Paulo
Abstract

Large language models trained on data generated by other models can sometimes inherit traits unrelated to that data, a phenomenon called subliminal learning. We study both hard label distillation, where students train on numbers generated by a biased teacher, and on-policy soft label distillation, where students generate the data and the teacher's logits provide the learning signal. Using Pythia, PolyPythia, and DataDecide OLMo models, we find that transfer persists in both settings across differences in weight initialization, pretraining data order, pretraining corpus, model size, and architecture. Across pretraining checkpoints, the ability to transmit a preference and the ability to acquire one develop separately: some checkpoints are more effective teachers but weaker students, and vice versa. Consistent with this distinction, cross-model transfer can be strongly asymmetric in magnitude when teacher and student roles are reversed. To investigate what carries the preference signal, we construct synthetic number sequences that match the teacher's generated number frequencies but have no dependence on the prompt or preceding numbers and find that these still induce substantial transfer in the hard-label setting. In top-$K$ soft-label distillation, removing animal tokens from the loss reduces transfer, although not completely in the same-model setting. In our experiments, soft label distillation produces more consistent transfer with less extensive training than hard label distillation. The growing use of on-policy distillation in model post-training makes this a particularly relevant setting for assessing unintended preference transfer.

0.28

DConstrain: Compiling Context-Parameterized Transitions for Diffusion Language Model Constrained Decoding

cs.AI ·2026-10-22 ·Shidong Cao, Jing Ma, Wenhao Chai, Yuchen Tian, Lun Du +1 more
Abstract · 39 pages, 6 figures, 12 tables

Diffusion language models (DLMs) decode multiple tokens in parallel, so individually valid tokens can be incompatible together under a schema. Exact joint inference over a grammar graph enforces compatibility, but it requires token-labeled successors across many hypothetical parser configurations. Encoding stacks and field histories in compiled states can cause exponential growth, which forces prior exact decoders to fix property orders and exclude valid outputs. We present DConstrain, which compiles schemas into context-parameterized transition programs. Cold compilation specializes token consumption to local grammar states while retaining context-dependent stack operations, environment guards, and caller-dependent deferred parsing symbolically. Tracking field histories as runtime values supports arbitrary property order without enumerating field subsets offline. At runtime, warm compilation binds reached configurations to these programs to construct exact successors, overlapping with the model forward pass because graph construction requires no model scores. Furthermore, token-effect grouping merges tokens with identical compiled effects before runtime contexts are bound, so groups are reused across stacks and environments even when concrete successors differ. Objective-specific score aggregation and token restoration provably preserve both the Viterbi optimum and the token-labeled path distribution of ungrouped inference. Across three benchmarks and two DLM families, DConstrain compiles 100% of BFCL schemas under arbitrary property orders, whereas an optimized finite-state baseline compiles only 71.30-85.81%. Under arbitrary property order, it reduces BFCL end-to-end decoding latency by 6.65-46.26 times with 100% schema validity and raises DiffusionGemma's BFCL accuracy from 37.27% under fixed order to 81.56%.

0.28

Is Complex Agentic Memory Always Needed? Naive RAG Is Stronger Than We Thought

cs.AI ·2026-10-22 ·Zhengye Bao, Shengtao Zhang, Haokai Zhang, Shiqi Lou, Muning Wen +1 more
benchmarkcommunity pick
Abstract · 62 pages, 5 figures. Includes appendices

LLM memory is widely used in long-term conversation understanding. Agentic memory systems often use mutation, structural organization, learned retrieval strategies, or other complex operations. Is this complexity always necessary? We study Raw-RAG, a naive RAG baseline using dense retrieval and reranking over raw history to supply evidence to a reader LLM. Across LoCoMo, LongMemEval, MEME, and MemoryAgentBench, Raw-RAG ranks first among the compared memory systems in six of seven benchmark-model settings, outperforming the strongest compared agentic memory system by 0.13-5.53 points. We investigate why raw-history retrieval remains competitive. Evidence analysis shows that its higher coverage of necessary evidence can offset lower utilization. Useful evidence includes both answer values and the relations and dependency paths needed to derive them. Even complete access to transformed memory stores leaves an accuracy gap relative to raw evidence, highlighting the importance of preserving raw information. These findings challenge the necessity of complex agentic memory for the evaluated conversation tasks and establish naive RAG as an essential baseline.

0.28

Consistent but Biased: Chair Personas Trade Noise for Bias in Multi-Agent Clinical Decisions

cs.MA ·2026-10-22 ·Yitong Bai, Chengyi Zhang, Ziyang Wang
Abstract · 32 pages, 8 figures

In clinical systems built from several large language model (LLM) agents, one agent often chairs the discussion and makes the final decision. Whether the chair's persona prompt changes that decision, or the agreement between repeated runs that some systems treat as reliability, has not been tested with the case held constant. We ask whether the persona changes the decision or only its timing, how it affects agreement between runs, and which behaviors appear when the prompt describes only personality traits. A simulated department of seven agents discussed 20 published cases of degenerative cervical myelopathy, with the case reports replayed unchanged in every run. Only the prompts of the chair and its deputy changed (no persona, or a consultative, autocratic or avoidant one), written either as rules of conduct or as personality traits. With personas written as rules, the chair chose to operate now in 93% of runs when autocratic and in 2% when avoidant, against 57% without a persona. Surgery was still planned in nearly all decisions, so the personas mainly changed whether to operate now or after more tests. These effects held with a second model and reworded prompts. The autocratic and avoidant personas made the three runs of a case agree more often, but only about as often as chance predicts, and Fleiss $κ$ did not rise. Prompts that described only personality traits kept more than half of the autocratic and avoidant effects on operating now. They reproduced quick decisions, delays and extra tests, but not ignoring red flags, dropping backup plans or passing the decision to the family. Studies should report and vary the chair's persona as they do the model, and correct agreement between runs for chance before calling it reliability.

0.28

What Comes Next and Why: Interpretable Next-Event Prediction with Neuro-Symbolic Rules

cs.LG stat.ML ·2026-10-22 ·Tim Nico Bauerschmidt, Isabel Valera, Jilles Vreeken
top venue
Abstract · Accepted at NeurIPS 2026

In real-world applications, being able to explain why a prediction is made is often as important - if not more so - than achieving high accuracy. However, existing methods for sequential event prediction typically prioritize accuracy at the expense of interpretability. Neuro-symbolic approaches offer a promising direction to improve interpretability by learning symbolic rules, but extending them to next-event prediction is challenging due to the presence of temporal dependencies. We introduce RULES, a neuro-symbolic model that learns sequential rules directly from data, and combine it with a residual network to form RUNES (Rule Network for Event Sequences). RULES accurately recovers ground truth rules from synthetic data and discovers compact, interpretable rule sets on real data. RUNES combines these rules additively with a residual network, achieving competitive accuracy while providing transparent access to active rules and how residuals adjust predictions.

0.28

AgentTriage: Replayable Security Diagnosis Across Heterogeneous Agent Runtimes

cs.AI cs.CR ·2026-10-22 ·Jie Ying, Xutao Zhang, Ningyuan Chen, Mengce Zheng, Jun Li +2 more
code/data
Abstract

When tool-use agents write files, execute code, or send data, their final responses do not explain the resulting security events, and heterogeneous runtimes expose the relevant evidence through incompatible interfaces. We present AgentTriage, an executable contract for replayable security diagnosis across heterogeneous runtimes. The contract admits conclusions whose claims cite canonical runtime events and requires abstention when mandatory evidence is missing. Runtime overlays normalize native observations, while a deterministic pipeline joins invocation and result lifecycles, constructs a typed provenance graph, selects a policy-decisive slice, and binds claims to recorded evidence. Given the same normalized trace, overlay version, and pre-run policy oracle, the diagnosis can be recomputed and its claims re-checked without language-model judgment. Across five runtimes, one contract captures all 159 applicable executions, with an audit of native and synthesized identifiers. On 120 adjudicated cases, AgentTriage achieves 0.858 grounded joint diagnosis, 0.966 required-evidence coverage, and a 0.036 unsupported-claim rate, improving joint diagnosis by 0.175 (95% CI [0.092, 0.258]) over the strongest same-schema baseline. These results quantify the precision of replayable diagnosis and the observation boundaries that require abstention. We release an open-source implementation and selected sanitized evidence at https://github.com/AsnGs/agenttriage-open-source.

0.28

Semi-Linear Gaussian Bayesian Networks: Enhancing Flexibility in Node Dependencies via penalized B-Splines

cs.AI cs.LG ·2026-10-22 ·Aitor Oviedo-Madrid, Mabel Morales-Otero, Jose Gonzalez-Gomariz, Ruben Armañanzas
Abstract · 25 pages, 6 figures

Bayesian networks for continuous variables often rely on restrictive assumptions: linear Gaussian models lack flexibility to capture nonlinear dependencies, while non-parametric approaches provide greater flexibility at a high computational cost. This creates a gap for modeling scenarios where variables are approximately Gaussian but exhibit nonlinear relationships. We introduce Semi-Linear Gaussian Bayesian Networks (SLGBN), a novel class of continuous Bayesian networks designed to bridge this gap, preserving Gaussian conditional distributions while introducing nonlinear conditional mean functions. The proposed framework provides a unified probabilistic framework combining linear Gaussian and nonlinear Gaussian conditional models within the same Bayesian network Nonlinear conditional mean functions are estimated using generalized additive models (GAMs) with penalized B-splines. We present parameter estimation procedures for both types of nodes, as well as a greedy hill-climbing algorithm for learning the network structure under the SLGBN framework. To assess its performance, we compare SLGBN with other continuous Bayesian network models, including semiparametric Bayesian networks (SPBN) and Linear Gaussian Bayesian networks (LGBN). Experimental results evaluate both goodness of fit and the ability to recover the underlying network structure. The results show that, in the presence of nonlinear Gaussian relationships, SLGBN achieves better fit and improved structure recovery compared to alternative models, particularly when combined with the Extended Bayesian Information Criterion (EBIC*) score. Furthermore, by retaining Gaussian conditional distributions, SLGBN offers computational advantages over fully semiparametric approaches.

0.28

IRON-Meme: A Benchmark for Structured Understanding of Identities and Roles in Online Meme Narratives

cs.AI ·2026-10-22 ·Gabriella Bollici, Stefano De Giorgis, Wouter Van Atteveldt, Filip Ilievski
benchmarkcode/data
Abstract · Submitted to ARR

As polarization grows on social media, understanding how political memes shape online narratives has become increasingly important. Previous research has relied on either qualitative frameworks that are hard to scale or computational methods that lack theoretical grounding. Both approaches treat narrative categories in isolation and focus exclusively on moral portrayals of identity. To address these limitations, we introduce a framework grounded in social science theories, encompassing three categories: social identities (such as gender, politics, and ethnicity), viewpoints (ingroup and outgroup perspectives), and narrative roles (moral character, intelligence, status, and power). Building on this framework, we present IRON-Meme, a human-annotated benchmark comprising 1,110 political memes across substantially more templates than existing benchmarks. Our analysis of IRON-Meme shows that most memes assign derogatory roles and that identities, roles, and templates are systematically interconnected. Evaluation of nine vision-language models indicates that frontier-scale and reasoning models achieve the highest performance, while accuracy on social identities and narrative roles decreases for less frequent labels. Incorporating template background knowledge does not enhance narrative role classification, suggesting that meme interpretation relies on how memes are adapted for specific communicative purposes. We make IRON-Meme and our code publicly available.

0.27

Governance-State Corruption in Long-Term Memory for LLM Agents: A Cross-Model Stress Test

cs.AI cs.SE ·2026-10-22 ·Bowen Liu, Youting Wang, Bonan Shen, Dingyan Shang
benchmark
Abstract · 13 pages, 1 figure, 10 tables

An LLM agent can retrieve a memory correctly yet use it incorrectly after its validity state changes. We study governance-state corruption: faults that alter which retained records enter model-visible context while stored content and tasks remain fixed. Across six providers, 731 benchmark-grounded bases yield 2,080 matched fault pairs. In offline control-plane checks, corruption changes the operative projection in 96.25% of pairs, while oracle-backed enforcement restores it in 100%. For downstream evaluation, a deterministically selected, non-probability LongMemEval/Memora panel resolves 298 of 300 matched triplets (238 distinct bases). Clean-minus-corrupted contrasts are 40.54 percentage points in LongMemEval binary correctness (95% base-cluster interval [30.32, 50.71]) and 30.60 points in rescaled Memora FAMA ([23.31, 37.69]); both sources also show recovery. OAKS Novel native option parity holds for all 6,714 resolved outputs. Full-cohort frozen-scorer corruption/recovery effects (-27.55/+27.51 points) remain descriptive proxies, not validated native semantic population effects. These bounded findings motivate evaluating memory control-state integrity alongside retrieval accuracy.

0.27

Phronix: The Capability-Centric Learning Paradigm for Agents

cs.AI ·2026-10-22 ·Hariharan Devarajan, Neeraj Rajesh, Aniruddha Marathe, Nikoli Dryden, Kin Ng +1 more
Abstract · 11 pages plus references

Scientific computing increasingly relies on large language model-based agents for performance engineering tasks such as instrumentation, analysis, and workflow porting. However, agent effectiveness depends heavily on the surrounding harness, which typically relies on manually written skills and curated memory. This limits autonomous learning, forcing agents to repeatedly rediscover solutions rather than retain and reuse knowledge from prior executions. We introduce Phronix, a capability-centric learning paradigm that enables agentic harnesses to acquire, validate, and refine reusable capabilities from their own executions. Phronix operates through a predictor-validator-corrector loop: MCP tools and associated skills propose solutions, tool executions validate them, and observed outcomes refine the stored capabilities. It distills every observed interaction into a capability store and researches a failing task rather than abandoning it. First, such learned capabilities solve a recurring task on the first attempt in 93-95% of cases across every model, significantly reducing solution non-determinism and improving reliability. Second, Phronix turns an agent's own execution into 7.6-30.8 durable capabilities per task without user instruction, and creation falls below two once the needed capability exists. Third, warm-start runs consume only 0.38-0.40 million tokens per task, which is three to four times less than a single baseline attempt, while a cold-start costs no more than the baseline. On the 1000 Genome Pegasus workflow, instrumenting a production application for I/O performance analysis, Phronix passes all 11 correctness checks against six for the baseline harness, and also lifts the task solving rates of two cheaper models from 29% and 27% of SWE-bench tasks solved to 91% and 87%, respectively, matching the performance of the most expensive evaluated model.

0.27

MOER: Mix-Policy Optimization under Expert Prefix Guidance with Selective Rollback

cs.LG cs.AI ·2026-10-22 ·Xuyang Zhong, Jiuding Yang, Yiwen Guo, Chen Liu
code/data
Abstract

Mix-policy optimization incorporates off-policy expert data into the reinforcement learning (RL) loop to transcend the capability boundary of on-policy RL. However, existing methods either provide static full-trace guidance that suppresses self-exploration, or adopt expert prefix guidance without correctness guarantees, leading to suboptimal performance or even training collapse. In this paper, we propose MOER, a mix-policy optimization framework that harmonizes expert guidance with self-exploration. MOER injects a decaying expert chain-of-thought prefix into the prompt and lets the model complete the remaining reasoning, with the prefix length progressively annealed to encourage self-exploration. A selective rollback mechanism replaces failed completions with the original full expert trace, guaranteeing the correctness of every off-policy trajectory. Furthermore, a harmonized mix-policy objective optimizes on-policy tokens with a GRPO loss while supervising both expert prefix and completion tokens with a trust-region SFT loss, substantially stabilizing training. Extensive experiments show that MOER achieves the highest average scores in mathematical and general domain reasoning tasks, outperforming all basic and mix-policy baselines, and generalizes consistently across diverse tasks, model scales, and base model categories. Codes are available at https://github.com/CityU-MLO/MOER.

0.27

Speculative Acceptance Does Not Decay with Position: A Full-Length Diagnostic on Long Chain-of-Thought with Independent Drafts

cs.AI ·2026-10-22 ·Zhiming Meng
hot direction
Abstract · 10 pages, 9 figures

The claim that speculative-decoding acceptance rate decays as a chain-of-thought grows longer has become a de facto industry narrative, motivating draft-model co-design and position-aware training. Nearly all direct evidence comes from trained drafters whose offline training length is far shorter than inference, so a purely positional law has never been isolated. We present the first systematic, full-length measurement of acceptance rate as a function of generation position on MATH-500 (all 500 problems), under greedy and sampling decoding, with two independent drafts against DeepSeek-R1-Distill-Qwen-7B: a same-family draft (DeepSeek-R1-Distill-Qwen-1.5B) and a cross-family draft (Qwen2.5-1.5B-Instruct), yielding about 1.78M verification rounds. Using within-problem paired tests, we find that after isolating a greedy-decoding repetition pathology, acceptance rate does not decay with position in any of the four configurations within the supported ranges (greedy 0-8K, sampling 0-12K); the second half of each trajectory is slightly but significantly higher than the first. A controlled temperature sweep shows that the apparent sampling-mode degradation of high-entropy reflection tokens is a visibility artifact of argmax agreement, not a drop in expected acceptance probability. Rejections concentrate with high precision on reflection trigger tokens such as "wait", "however" and "check", an entropy-selective failure that a lightweight three-arm LoRA trajectory-alignment intervention (uniform, reflection-marked, entropy-weighted) fails to improve in mean accepted length. We thus supply the positional baseline that independent draft-target pairs were missing, and localize the residual failure to decision-point entropy rather than position.

0.27

ProQuant: Preserving Expert Quality over Quantity in Ultra-Low-Bit Quantization of Mixture-of-Experts Models

cs.AI ·2026-10-22 ·Jiaqi Chen, Xianglong Yan, Chengzhu Bao, Xie Chen, Yulun Zhang
code/data
Abstract

Mixture-of-Experts (MoE) large language models reduce computation through sparse activation but still incur substantial storage costs. Mixed-precision post-training quantization (PTQ) reduces these costs by quantizing different model components at different bit widths, but existing methods often struggle to preserve generation quality under tight bit budgets. We identify two key challenges: (1) nonlinear quantization errors at extremely low precision are underestimated when cross-layer propagation is ignored; (2) direct router fine-tuning may cause routing drift, and calibration gains may not generalize to downstream tasks. To address these issues, we propose ProQuant, a weight-only PTQ framework for MoE models. Prune-or-Prioritize Mixed-Precision Allocation (PoP) jointly optimizes expert pruning and bit-width assignment. It allows higher-precision experts to substitute for severely degraded ones while improving bit-budget utilization. Routing-Consistent RMSNorm Finetune (RoCo) adapts expert input representations through mathematically equivalent transformations. It compensates for quantization errors to reduce routing drift. Extensive experiments demonstrate that ProQuant outperforms existing methods across diverse generative tasks. Code is available at https://github.com/ThrowBug/ProQuant.git.

0.27

An Agent Architecture with Tools and Memory for Selective Correction of Wrist PPG Heart Rate Estimates

cs.LG ·2026-10-22 ·Xu Wang, Shuhan Yi, Zhen Cheng
Abstract · 71 pages, including 48 pages of Supplementary Information

Wrist photoplethysmography (PPG) heart rate estimates are vulnerable to motion artifacts, and corrections that reduce mean error can worsen individual windows. We propose an agent architecture with tools and memory around a fixed heart rate estimator. Pulse, motion, and limited historical evidence generate candidate estimates, whose predicted benefits and harms guide selection. An independent risk gate determines whether to retain, correct, or reject each output. We trained the architecture on PPG-DaLiA and tested it on Wrist Exercise, GalaxyPPG, and WildPPG with frozen weights. Across these four public datasets, mean absolute error (MAE), macro-averaged over ten estimators, decreased by up to 44.5%, with the largest reduction on GalaxyPPG. Beneficial and harmful corrections decreased and increased absolute error by more than 1 bpm, respectively. Their ten-estimator macro-averaged proportions ranged from 16.06% to 36.64% and from 1.68% to 6.77%, respectively, across the four datasets. At 60% and 80% target coverage, numerical correction on the same accepted windows contributed more to total error reduction than window selection. We implemented real-time wearable acquisition, Bluetooth reception, and smartphone deployment of a selected estimator and the framework. In backend analysis of four prototype recordings from one participant, ten-estimator macro-averaged MAE decreased from 9.09 to 4.45 bpm without additional rejection. Under the evaluated conditions, the architecture improved average accuracy while quantifying correction benefits and harms, supporting wearable heart rate analysis that considers error, correction risk, and coverage.

0.27

LLM Multi-Agent Systems Are Vulnerable to Herding and Runaway Cascades

cs.AI cs.MA ·2026-10-22 ·Farnoosh Hashemi, Shiyu Ji, Joice Chen, Michael W. Macy
Abstract

Large language models (LLMs) increasingly operate in multi-agent systems where they collaborate by responding to others' choices, creating endogenous feedback loops with vulnerability to social influence, herding behavior and runaway cascades. Following the design of the classic Music Lab experiment, we tested this vulnerability in large populations of LLM agents that recommended unfamiliar songs, with or without exposure to cumulative recommendation counts. Exposure led to runaway cascades that generated extreme and unpredictable collective recommendations increasingly divergent from a song's intrinsic appeal. Compared to humans, LLM populations were generally far more susceptible to herding behavior, and unlike humans, LLM cascades were less able to self-correct. This vulnerability has implications for multi-agent collaboration in recommender systems, automated evaluation, complex problem-solving, and unauthorized conspiracies.

0.27

Effectiveness of Curriculum Learning Depends on Reward Sparsity and Competing Optima

cs.LG cs.AI ·2026-10-22 ·John J. Vastola, Ann Huang, Satpreet H. Singh, Samuel J. Gershman, Kanaka Rajan
top venue
Abstract · Accepted to NeurIPS 2026

While animals and people tend to learn a task more quickly and reliably when they first train on simpler versions of it, curriculum learning's effectiveness in artificial settings varies, and appears substantially greater in reinforcement learning than in supervised learning. In this work, we construct and analyze a minimal model of policy learning to better understand why. We consider an open-loop, episodic target interception task whose parameters---especially how close the agent must be to the target to receive an informative reward signal---can be chosen so that the agent either receives informative rewards frequently ('dense' rewards) or rarely ('sparse' rewards). We mathematically and numerically analyze the dynamics of REINFORCE-based policy learning in both cases. In the dense reward case, curricula have only marginal benefits; in the sparse case, curricula can be required for learning to occur at all. We find that this is especially true when the agent has the conflicting goals of minimizing effort and accumulating high reward, since this can produce distinct reward landscape optima which compete with one another. Our characterization of useful curricula in this setting indicates that they address two issues: they (i) make rewards less sparse to speed up learning, and (ii) shape the reward landscape to steer agents away from bad local optima.

0.27

Doubly Robust Distributionally Robust Offline Contextual Pricing

cs.LG ·2026-10-22 ·Min Xu, Xinyi Yin, Yunfan Zhang, Yuxuan Han, Houcai Shen +1 more
top venue
Abstract

Offline contextual pricing often relies on logged observational data, but faces challenges from distributional shifts between training and deployment environments. Distributionally robust optimization (DRO) provides a principled approach to off-policy evaluation and learning (OPE/L). However, existing methods are mostly limited to discrete actions. Recent work has explored DRO for continuous treatments using inverse propensity weighting (IPW), while such IPW-based estimators can be sensitive to the convergence rate of propensity score estimates, particularly when estimated nonparametrically, which may lead to larger estimation errors and regret. In this work, we develop a doubly robust (DR) framework for distributionally robust OPE/L in continuous pricing settings. For evaluation, we propose a localized DR estimator that addresses the computational challenges of worst-case expectations by fitting only a small number of regressions, comparable to standard non-robust DR, while achieving semiparametric efficiency under mild product rate conditions. For learning, we leverage the inherent smoothness of demand noise to handle pricing-specific discontinuities in revenue outcomes (e.g., threshold-based purchase decisions), establishing a finite-sample regret bound of $\tilde{\mathcal{O}}_p(T^{-s/(2s+1)})$ for smoothness orders $s=1,2$. This bound improves upon existing regret rates in existing DRO-based off-policy learning (OPL) for continuous treatments. Extensive experiments under various levels of distribution shift validate our proposed framework.

0.27

Conservative Meta-Agent Search: Architecture Adaptation under Non-Stationarity

cs.AI cs.LGcs.MA ·2026-10-22 ·Ali Atiah Alzahrani
benchmark
Abstract · 24 pages, 8 figures, 17 tables

Automated agentic-system design is typically evaluated on static task suites with immediate scalar rewards. Deployed systems face persistent shift, delayed dependent feedback, correlated component failures, and costly architecture changes. We formalize meta-agent architecture search as sequential incumbent replacement with switching costs and hard risk constraints and introduce Conservative Meta-Agent Search (CMAS). CMAS retrieves archived architectures and proposes typed graph edits, then scores each against the incumbent by paired shadow replay under common random numbers, with interventional component credit. Promotion requires an empirically calibrated, variance-penalized simultaneous score above a pre-specified margin, and is followed by a simulated canary stage and pre-specified rollback. Since backtests cannot certify meta-level skill, we introduce MetaInvest-Bench, a regime-switching simulator whose specified worker competence makes a surrogate-optimal architecture computable, with leakage-controlled replay, adversarial governance tracks, and five negative controls. Financial streams serve as a risk-sensitive testbed for meta-level generalization; we claim no alpha. Across 96 streams, CMAS cuts normalized mean surrogate dynamic-oracle regret from 0.116 to 0.104 and raises held-out utility from 0.44 to 0.48 over the strongest adaptive baseline; when extra workers add no value, it contracts to near-single-agent teams. Its deployment inference cost is ~6% lower, though its search-inclusive token cost is the highest of the methods compared. The observed per-update false-promotion rate is 3.4% against a nominal 5% target, and 4.8% in a sealed rerun. The calibration is empirical: it concerns an update-level replay statistic on an author-designed benchmark, and guarantees neither full-utility improvement, nor absence of simulated canary-stage harm, nor sequence-level error control.

0.26

Ground-Truth-First Benchmarking of Reinforcement Learning via Converse Optimality: From Smooth, Fully Observed Systems to Non-Smooth or Partially Observed Tasks

cs.LG ·2026-10-22 ·Grégoire Ouerdane, Sinan Ibrahim, Stefan Streif, Pavel Osinenko
code/data
Abstract

Benchmarking in reinforcement learning is usually relative: algorithms are ranked by returns on environments with unknown optima. Distance to optimality remains unobservable, while conclusions may depend on reward design, stochasticity, hyperparameters, environment selection, and simulator details. We develop a ground-truth-first framework based on converse optimality, constructing each task with a certified optimal value function and policy withheld from learners and used only for evaluation. The framework unifies two modes: sensitivity benchmarking samples certified environments from continuous-time generators to assess robustness to task variation, while oracle-referenced comparison uses native discrete-time generators to evaluate policies via optimality gap, regret, and policy error. Continuous-time models support physically interpretable sensitivity studies but introduce dependence on the simulator step ($Δt$) and numerical scheme. Native discrete-time models align the environment, oracle, and learner on the same transition law, eliminating external integration choices as a confounder. We reproduce quantitative results covering optimizer sensitivity and oracle-referenced comparisons on serial planar arms, dynamically extended nonholonomic vehicles, and quadrotor-inspired tasks. The methodology extends beyond smooth affine Gaussian systems: virtual-input pullback handles non-affine raw-action interfaces and finite-covariance non-Gaussian disturbances. A non-smooth converse optimality formulation enables certified benchmarks with kinked value functions and switching surfaces. For partially observed evaluation, an information-state formulation reduces benchmarking to converse optimality on a sufficient statistic and provides an information-penalty reference. Code, fixtures, and diagnostics are available on the public project page.

0.26

RheOPT: Few-Shot Non-Newtonian Flow Prediction via Newtonian Pretraining of RheOFormer

cs.LG ·2026-10-22 ·Maedeh Saberi, Amir Barati Farimani, Safa Jamali
Abstract · 15 pages, 7 figures, Supplementary Information

Despite advances in numerical methods, non-Newtonian flow simulations remain computationally expensive because velocity and stress must be advanced together through rate- and history-dependent constitutive dynamics. Neural-operator surrogates reduce inference cost after training, but their training data require the same expensive simulations they aim to replace, making data-efficient learning essential when only a few non-Newtonian simulations can be afforded. Here, we introduce the Rheological Operator Pretrained Transformer (RheOPT), which pretrains a transformer-based neural operator on inexpensive Newtonian simulations and fine-tunes it on a small number of non-Newtonian samples. Rather than transferring the Newtonian constitutive response itself, RheOPT reuses geometry-conditioned transport structure that remains useful after the stress model changes. We evaluate RheOPT in two geometry-matched settings: a $4{:}1$ contraction channel for Giesekus, Oldroyd-B, and Bingham fluids, and the oscillatory wake behind a triangular obstacle for Oldroyd-B flow. In the contraction, Newtonian pretraining substantially improves few-shot prediction, and a model fine-tuned on four target samples can match or exceed a from-scratch model trained on four times as many. This benefit is reliable under reversible instance normalization (RevIN) but dataset dependent under Global normalization, a difference we trace to the dynamic range of each fluid's velocity and stress fields. In the wake, where relative $L_2$ error alone cannot verify that vortex shedding is recovered, phase-insensitive diagnostics show that pretraining recovers the shedding frequency and wake fluctuation amplitude that are suppressed by the from-scratch model in the few-shot regime. Overall, RheOPT establishes Newtonian simulation as a physically grounded pretraining source for data-efficient non-Newtonian operator learning.

0.26

OpenAttention: One Percent Is Enough with Tail Proxies

cs.LG cs.AI ·2026-10-22 ·Hyunwoo Yu, Jongbeom Lee, Hyunbin Kim, Sungjun Park, Jaemin Choi +6 more
code/data
Abstract · 27 pages. v2: revised Appendix C and corrected abstract wording. v3: corrected author and affiliation formatting in the HTML version; scient…

Sparse attention reads only a small subset of the key-value cache, and how to treat the unread tail remains a key challenge. Discarding the tail biases the output, and sampling it as previous approaches do fails at reads of 5% and below, where the tail likely holds heavy hitters the selector missed. Our key observation is that the simplest possible proxy, one token that folds the entire tail into its count-aware mean, already yields lower output error than a sample at every read behind a practical selector. We further show in closed form that at low reads sampling nearly doubles the weight error of discarding while a single proxy stays below it. Building on these findings, we explore approximating the dense attention distribution with a few proxies that target the tail. We propose OpenAttention, a pipeline of clustering, ErrorGate and top-k selection followed by a count-aware mean proxy. To implement clustering and selection both efficiently and effectively, we design training-free Query-Geometry Clustering (QGC) and a Query-Geometry Selector (QGS), which compress the keys while respecting the query directions. Additionally, unlike prior methods that rank clusters by centroid alone, ErrorGate allocates the exact read by the estimated weight error of each cluster. On RULER-32K HARD with Llama-3.1-8B, OpenAttention retains 99% of dense accuracy at a 1% read. It also achieves state-of-the-art accuracy across models, context lengths and datasets. Our customized attention kernel for OpenAttention runs the decode attention stage 10$\times$ faster than dense attention at a 1% read.

0.26

Who Is Sycophantic to Whom? Confident LLMs, Human Deference, and the Collapse of Trust Calibration

cs.AI cs.HC ·2026-10-22 ·Deepthika S, Pawni Yadav, B. Sankar
Abstract · 49 pages; 32 figures; 7 tables

Research on sycophancy asks whether large language models (LLMs) bend towards what their users want to hear. We ask the reverse question: when an LLM is confident, consistent, and wrong, do users bend towards it? In a within-subjects experiment, 32 participants learned up to five science topics by chatting with five assistants built on the same model (GPT-5-mini). Unknown to participants, each assistant's briefing had 0, 25, 50, 75, or 100% of its facts replaced by plausible falsehoods, and every assistant spoke with full confidence. A surprise quiz after each topic offered the correct fact, the planted falsehood, and two distractors, with a confidence rating for every answer. Across 104 quizzes and 1,040 answers, scores fell from 7.4 to 2.9 out of 10, and the planted option was chosen on 60% of falsified items against 11% of correctly taught ones (odds ratio 12.5). Transcript coding showed that whenever the assistant stated a falsehood, it was adopted about 65% of the time at every falsification level. Absorbed falsehoods were held with high confidence (8.4/10), the modeled confidence advantage of right over wrong answers vanished, and the discrimination of confidence for correctness fell from an AUROC of 0.70 to chance, while post-task trust ratings barely moved. Participants tended to return to the chat more when the assistant was worse, but returning went with more absorption, not less, and 30% of all turns pasted quiz questions back to the source being checked. The assistant itself was almost never sycophantic and never conceded under pushback. We introduce a baseline-corrected Human Deference Index to quantify this human-side sycophancy and argue that trustworthiness is a property of the human-AI pair, that sycophancy evaluations should measure both directions, and that learning interfaces must make reliability legible claim by claim.

0.26

ValuSpec: Plug-and-Play Candidate Valuation before Target Verification for Tree-Based Speculative Decoding

cs.AI ·2026-10-22 ·Liang He, SiYuan Ma, Qishi Zhan, Yongqi Fan, Mingyu Cao +3 more
published
Abstract · NeurIPS 2026 poster

Tree-based speculative decoding accelerates large language model inference by verifying multiple candidate branches in parallel. However, in practical serving scenarios, similar requests repeatedly induce overlapping candidate branches. While retrieval-based methods mitigate this by reusing historical drafts, the resulting candidate trees often contain many low-quality branches as the candidate tree grows, which reduces verification efficiency and wastes target-side computation. To address this, we propose ValuSpec, a plug-and-play candidate valuation mechanism that filters low-quality branches from a supplied candidate tree before expensive target-side verification. ValuSpec introduces transfer-based supervision to align filtering with the target model's behavior. A filtering threshold is estimated from the target model's greedy trajectories on a calibration split, without any parameter training. The target model then verifies the filtered tree under the output-identical greedy rule, so accepted tokens are guaranteed to match standard autoregressive decoding exactly. Experimental results on HumanEval, GSM8K, Dolly, and SpecBench show that, in our main evaluation, ValuSpec achieves up to 2.38x end-to-end speedup in retrieval-based tree settings. Furthermore, integrating ValuSpec into the EAGLE-3 dynamic tree framework yields up to 2.78x end-to-end speedup without modifying its tree construction strategy, demonstrating that ValuSpec serves as a generally applicable and plug-and-play candidate valuation module.

0.26

Towards Field-Deployable Wearable IMU Systems for Occupational Activity Recognition

cs.LG ·2026-10-22 ·Ihsan Ul Haq, James Brusey, Farnad Nasirzadeh, Chandan Karmakar
publishedbenchmark
Abstract · Accepted at IEEE BSN 2026

Wearable inertial measurement unit (IMU) systems for construction worker human activity recognition (HAR) routinely report aggregate accuracies above 95 %, rarely under genuine subject-independent (leave-one-subject-out (LOSO)) evaluation, masking the per-individual variance that determines real-world deployment performance. Using the VTT-ConIoT benchmark, we conduct, to our knowledge, the first systematic configuration sensitivity analysis of this dataset: six classifiers across five sensor placements and three window sizes under a single LOSO protocol, a personalised-calibration pilot, and a validated comparison of tabular GAN-based and standard time-series augmentation. Per-subject LOSO accuracy for the main multi-sensor, 16-class configuration ranges from 61.9 % to 88.7 % (mean 77.5 %), a 27 pp spread, though all data are laboratory-collected. Multi-sensor fusion and longer windows both improve accuracy; standard augmentation gives small, consistent gains, while Conditional Tabular GAN (CTGAN) augmentation actively harms accuracy with a demonstrated distributional-fidelity mechanism; and personalised calibration recovers 16.8 pp for the weakest participants from just 30 % of their own data. These findings provide evidence-grounded configuration guidance for scaling occupational body sensor networks towards field deployment

0.26

Mitigating Saliency Collapse: Robust Saliency-Aware Long-Text Image-Text Alignment

cs.AI ·2026-10-22 ·Qiuyu Kong, Zanxi Ruan, Marco Cristani, Yiming Wang
code/data
Abstract

Long-text image-text alignment is essential for fine-grained scene understanding, yet existing methods mainly focus on extending context length or improving local alignment. We identify an overlooked failure mode, saliency collapse, where models over-rely on text that features visual prominence while underutilizing non-salient context. Consequently, retrieval performance degrades sharply when salient content is reduced, despite that the non-salient context remains informative. To address this, we propose RoSA, a saliency-aware finetuning framework that decomposes image-text pairs into complementary salient and contextual views, enabling joint learning of prominent and contextual semantics. We propose a modality-independent decomposition strategy to combine object-level visual saliency estimation with LLM-based text decomposition to bridge vision and language. We employ auxiliary objectives to explicitly align these views, ensuring robust representation with both salient and contextual features. RoSA greatly improves long-text retrieval and demonstrates strong robustness to saliency collapse across benchmarks. Moreover, our auxiliary objectives are plug-and-play, bringing consistent gains to existing tuning methods. Code and checkpoints are available at: https://github.com/intelligolabs/RoSA.

0.26

Inverse Modeling for Laser Pulse Shape Design in Inertial Confinement Fusion

cs.AI cs.LGeess.SYhep-ex ·2026-10-22 ·Ricardo Luna Gutierrez, Vineet Gundecha, Rahman Ejaz, Varchas Gopalaswamy, Riccardo Betti +3 more
Abstract · Initial version submitted to NeurIPS 2026 pending publication

The achievement of practical fusion energy remains one of the most pressing unsolved scientific challenges, with immense implications for carbon-free power. A critical factor in the success of Inertial Confinement Fusion (ICF) is the design of a Laser Pulse Shape (LP) that can optimally drive implosions under stringent physical constraints. Traditional LP design relies on computationally expensive simulations and manual iterative refinement. We introduce the ICF Laser Pulse Shape Design System (LPDS), a generative inverse modeling framework that maps desired outcomes and target pellet configurations directly to optimized LPs. Crucially, we design a multi-objective loss function to ensure the generated LPs satisfy fundamental physical constraints and experimental requirements. Furthermore, we present constraint-conditioning, inpainting, and gradient-based LP editing mechanisms for fine-grained control over specific pulse characteristics during generation. Moreover, we study the transfer of our framework to a limited set of real experimental shots. Our method establishes a data-driven inverse design framework for LP design in ICF, contributing to the advancement of practical and sustainable fusion energy.

0.26

Conflict-Aware Logit Adapters for Utility-Preserving Anti-Distillation

cs.AI ·2026-10-22 ·Quoc Phong Dao, Binh Anh Nguyen, Linh Ngo Van, Thanh-Toan Do, Mehrtash Harandi +1 more
published
Abstract · NeurIPS 2026

While reasoning traces improve model capability, they also provide supervision signals for downstream distillation. Initial efforts to mitigate this issue rely on decoding-time interventions, but they incur high inference overhead and can harm the teacher model's natural fluency. We propose Conflict-Aware Logit Adapters (CALA), a lightweight training-time defense. CALA attaches a small residual adapter to a frozen teacher model and trains only the adapter. A proxy student identifies positions where imitation is most sensitive, and the adapter perturbs the output distribution at these positions while remaining within high-probability regions. A conflict-aware optimization strategy balances anti-distillation and utility objectives, while regularization preserves the teacher's performance and fluency. At inference time, CALA incurs modest overhead and requires no auxiliary model. Experiments on reasoning benchmarks show that CALA substantially reduces successful student imitation while maintaining near-original teacher performance and generation quality. These results establish CALA as a practical and efficient alternative to decoding-time approaches.

0.26

Early learning in large residual networks

cs.LG stat.MLmath.PRphysics.data-an ·2026-10-22 ·Vladimir Narovlansky, Boris Hanin
Abstract · 23 pages + appendices, 10 figures

We study the early training dynamics of nonlinear residual networks with blocks of zero hidden layers as a model of more general residual networks. We analyze networks in a sequential limit where first width and depth diverge, then the number of training data points diverges, and finally the input dimension diverges. This is a more refined setup compared to the typical mean-field theory analysis in which depth and width diverge but the number of training datapoints and input dimension are fixed. Our main technical result is a diagrammatic computation of both the network output and the hidden representations after the few steps of gradient descent in terms of the statistics of inner products of inputs from the training set. One of our principle conceptual findings is a negative result: in the limit of large input dimension the network output can learn only a linear function of its inputs early in training since higher learned monomials are more and more suppressed in the input dimension and vanish in the large dimension limit. This suggests that mean-field limits of ResNets are surprisingly rigid in the regime of high input dimension and may not be sufficient toy models for understanding feature learning. Finally, we investigate how to ensure hyperparameter transfer for the learning rate as a function of dataset size and input dimension. We test our predictions numerically and find excellent agreement with our theory.

0.26

DeepOpen: Decision-Oriented Intent Classification on BANKING77 and CLINC150

cs.AI ·2026-10-22 ·Shuang Song, Muhammad Luqman Naseem, Zhongqiao Yuan
Abstract

We present DeepOpen, a decision-Oriented intent-classification framework with dataset- specific systems for BANKING77 and CLINC150. On BANKING77, a single English classi- fier continued with sharpness-aware minimization (SAM) achieves 94.1234% test accuracy (2,899/3,080). This is the best observed result among three fixed SAM seeds, selected for display after test evaluation; their mean is 94.0368%. On the 150 in-scope CLINC150 classes, probability fusion of one Laya specialist and three DeBERTa-v3-large classifiers achieves 98.0222% (4,411/4,500). We describe the two configurations, reproduce the reference com- parisons displayed in the accompanying evaluation explorer, and separate headline scores from evidence about improvement. Under a uniform BANKING77 evaluation path, the SAM mean exceeds the historical checkpoint by 0.0433 percentage points, with a shared-row 95% interval of [-0.2273, +0.3247]. Thus, the continuation does not establish a reliable accuracy gain. The report documents strong observed task-specific performance, selection effects, and unresolved historical execution-path differences, rather than a new backbone or a unified algorithmic advantage.

0.26

Learning Structured Sparsity for Large Models: Bridging Model Compressibility and Efficient Inference

cs.LG cs.AI ·2026-10-22 ·Tianxiang Xia, Kai Lion, Niao He, Michael Muehlebach, Melanie Zeilinger +1 more
Abstract · 35 pages, 26 figures, 14 tables. Project website: https://egretwalker.github.io/S-SALAAD/

Scaling laws continue to reward increases in model scale, while the resulting computational and memory demands drive greater reliance on cloud infrastructure for inference. Yet unreliable network connectivity, communication delays, and privacy requirements can make such reliance impractical, motivating local execution on edge devices. Local deployment requires models to fit within limited memory and execute efficiently with the available computational resources. Recent work has shown that learning sparse-plus-low-rank representations enables elastic compression of a single checkpoint to multiple memory budgets. However, fewer parameters do not necessarily translate into faster inference. Indexing overhead and irregular memory access in unstructured sparse components can offset computational savings. To address this critical deployment barrier, we introduce S-SALAAD, a unified framework that jointly learns the model weights and their low-rank and structured sparse components during training. The framework supports a family of hardware-compatible structured sparsity patterns. It couples this structure learning with budget-aware compression and dedicated CPU and GPU kernels that exploit the learned structure for acceleration, enabling efficient inference from a single checkpoint across memory budgets. Experiments on Llama2 and Qwen3 models, covering both pretraining and distillation, show that S-SALAAD advances the empirical Pareto frontier in perplexity, memory, and inference latency across multiple hardware platforms relative to the evaluated baselines. These results show that structured learning can translate elastic compressibility into reduced memory requirements and lower inference latency, enabling practical deployment of large models on edge devices.

0.25

Reject, Remember, Retry: Why Multi-Turn Attackers Should Remember Their Mistakes

cs.AI ·2026-10-22 ·Jekaterina Novikova, Yuqing Huang, Jack Ji, Sagarika Suresh, Jithin Pradeep +1 more
Abstract · v2 is a substantially revised and extended version, retitled "Reject, Remember, Retry" ($R^3$). v1 was presented at the AdvML-Frontiers x Co…

Existing multi-turn jailbreak attackers condition on the same conversation history as the target. We introduce $R^3$ (Reject, Remember, Retry), a black-box multi-LLM framework in which the attacker conditions on a compact state rather than the raw transcript. Turns that fail to raise the attack risk score are removed from the target's context but remembered by the attacker, together with the reason for failure. We call this asymmetric memory. Across two benchmarks and six target LLMs, $R^3$ outperforms state-of-the-art baselines by up to 53% in ASR@5 under matched query budgets, and a low-capacity attacker with $R^3$'s state is able to outperform high-capacity attackers without it by up to 34% in ASR@5. Ablations attribute most gains to remembering rejected turns, and the risk score acts as a coarse threshold rather than a progress measure.

0.25

Cognitively Constrained Partners for Zero-Shot Coordination in Real-Time Partitioned Control

cs.LG cs.HC ·2026-10-22 ·Ash Simmons, George De Ath, Richard M. Everson
published
Abstract · 30 pages, 6 figures, 5 tables. Accepted at the NeurIPS 2026 Workshop on Human-AI Coevolution: Measuring Human-Agent Teams in the Agentic Era

Effective human-machine teaming requires agents that coordinate with unpredictable human partners. Reinforcement learning from live human feedback or large behavioural datasets is expensive, while purely simulated training overfits to artificial conventions and fails at Zero-Shot Coordination with real people. We propose a framework for training teaming agents against Cognitively Constrained Partners: simulated partners whose diversity emerges from procedural constraints grounded in human physiological literature, specifically reaction delays and perceptual limitations, rather than from reward shaping or historical training checkpoints. In a modified Lunar Lander environment, we report preliminary evidence that these partners play more like real people than unconstrained agents do, and that agents trained against them match the strongest cross-play baselines while being the only ones whose performance does not depend on how their training partners were generated. We also describe a human-participant study designed so that its rankings can be compared directly against cross-play, testing how well that standard evaluation predicts real human-agent teaming.

0.25

Multichoice Modeling: Exploring Decision Configuration Space under Uncertainty

cs.AI q-fin.RMstat.CO ·2026-10-22 ·Mariia Kozlova, Roman Stepanov, Alexander Walzer, Andrea Saltelli
unusual
Abstract

Decision support has become sophisticated at exploring uncertainty, but much less so at representing the architecture of the decision itself. As a result, the decision space is often narrowed before uncertainty is analyzed, leaving interactions among choices and the effects of alternatives they imply largely invisible. This paper introduces multichoice modeling, a framework for making large decision configuration spaces explicit and computationally explorable under uncertainty. Drawing on the logic of rugged landscapes, we conceptualize such problems as contingent performance landscapes, where interdependent choices generate configuration-specific outcomes whose value changes across uncertain futures. The computational framework combines multidimensional decision and uncertainty spaces in a single simulation. Further, variance-based sensitivity analysis is used for identifying influential choices, and conditional distribution analysis for comparing configuration-specific performance profiles. We demonstrate the framework through three case studies: a manufacturing process upgrade, M&A strategy, and mortgage design. By contrasting conventional uncertainty-only analysis with multichoice modeling, the cases show how opening the decision configuration space can reveal unexpected complementarities, near-best configurations with different risk profiles, and strategies that were not apparent as alternatives at the outset.

0.25

Is spectral predictability enough? Testing Ω-matching for synthetic pretraining under distribution shift

cs.LG ·2026-10-22 ·Edric Castel C. Hao, Juan Raphael S. Seña, Neon S. De Jesus
benchmark
Abstract · Submitted to ICASSP 2027

Synthetic pretraining under distribution shift requires deciding what synthetic data to use. A natural criterion is to match the spectral predictability $Ω$ of the source to the target, but is $Ω$ sufficient? We test this hypothesis, which we call predictability-aware pretraining, across three architectures (LSTM, PatchTST, TCN) and three seeds. Under ideal synthetic-to-synthetic conditions, $Ω$-matched pretraining can outperform even an oracle baseline (+46.8% MSE reduction, LSTM), though the benefit is already architecture-dependent. On the real-world ETTm1 benchmark, however, this relationship does not hold: sine-plus-noise pretraining at matched $Ω$ yields only a marginal mean improvement +2.3% averaged over all three architectures, up to +65.5% on individual channels), whereas $Ω$ matched chaotic ODE trajectories systematically degrade performance -13.2%). The two sources agree on the direction of transfer for only 53% of configurations across all three architectures (Pearson r=-0.02), indicating that $Ω$ carries little information about which source will transfer beneficially. We further observe an architecture-dependent asymmetry in the response to the direction of the $Ω$ gap. We conclude that $Ω$ matching can help in controlled settings but is not sufficient to predict transfer across sources, and we discuss additional signal properties that a source-selection criterion may need to consider.

0.25

Learning Domain Trajectories with Flow Matching for Gradual Domain Adaptation

cs.LG ·2026-10-22 ·Yubo Huang, Yushe Cao, Zixi Wang, Jingzehua Xu, Xin Lai +2 more
top venue
Abstract · Accepted by NeurIPS 2026

Large distribution shifts remain a primary obstacle for domain generalization and unsupervised adaptation. Although gradual domain adaptation alleviates this difficulty by introducing ordered intermediate domains, most existing approaches still operate in a discrete, step-wise manner, which can accumulate errors and often relies on unstable Min--Max optimization. We propose \textbf{FlowGDA}, a flow-matching framework that models gradual adaptation as \emph{continuous} transport on a statistical manifold in a representation space. FlowGDA learns time-dependent velocity fields that induce smooth domain trajectories, using a simulation-free regression objective rather than adversarial training. To prevent semantic drift, we couple left-to-right transport between consecutive domains with a source-anchored transport and a semantic consistency loss, and we further stabilize learning via a two-stage sequential optimization scheme. Extensive experiments demonstrate that our method outperforms state-of-the-art techniques, validating flow matching as an effective alternative for domain adaptation.

0.25

Evaluating Artificial Wisdom (AW): A Trajectory-Based Framework for Context-Sensitive Decision Regulation

cs.AI ·2026-10-22 ·Md. Jalil Piran
Abstract

Capability benchmarks measure what large language models can do, but not whether their recommendations remain appropriate as decision conditions change. This paper proposes a trajectory-based measurement framework and prospective validation protocol for artificial wisdom, defined behaviorally as context-sensitive regulation of judgment. Versioned reference specifications determine admissible responses and transition materiality before evaluation. Seven stage-level criteria assess judgment under current conditions, while two transition components assess material-update adequacy and commitment-preserving stability. Stability requires prior adequacy; switching within registered commitment classes is permitted, and persistent error earns no stability credit. A nine-component profile is primary, with a secondary bottleneck score and separate critical-failure reporting. Three simplified simulations examine precision, bottleneck estimation, and sensitivity to model-family count. A single-criterion simulation met its precision criteria at 24 scenarios per domain. Near a bottleneck tie, the estimated minimum had a downward bias of 0.55 points at that size. A simplified incremental-prediction simulation detected a Brier improvement of 0.027 with probabilities of .508 and .742 using eight and twelve families, respectively; it does not establish power for the operational protocol. An exploratory predecessor-rubric study comprising 71 evaluation records illustrates aggregation sensitivity but does not validate the proposed instrument. The twelve-family target remains provisional pending feasibility assessment and calibration of the complete analysis. Whether the profile predicts independent decision outcomes beyond capability remains an open empirical question.

0.25

Joules-to-GDP: How Much Economic Value Do AI Systems Produce per Unit of Energy?

cs.LG econ.GN ·2026-10-22 ·Amirreza Zeinali, Jon Saad-Falcon, Avanika Narayan, Mersad Abbasi, Jonathan S. Hartley +3 more
Abstract

AI systems are increasingly used to complete economically valuable tasks in various professions, such as software engineering and finance. This motivates the question: how much economic value does an AI system produce per unit of energy? To answer it, we introduce Joules-to-GDP, a framework that measures output dollar value minus AI execution cost per unit of energy, which we call Net Value per Joule (NVJ), and expert time saved per unit of energy, or Time Saving per Joule (TSJ). We then apply the framework to 220 GDPval tasks across 44 occupations. Four findings stand out. First, the most valuable work is not always the most efficient. For example, engineering work such as CAD design yields about 10% more net value per task than software development such as building websites, yet the latter yields about 27% more per joule. Second, defaulting to the most capable model can mean less total value from the same energy budget. Of the five tested models, GLM 5.2 delivers the most net value, $72.27 per task on average, using the energy of about four hours of active laptop use, while Gemma 4 31B achieves 56% of that with 23% of the energy, for 2.5x the NVJ and 2.8x the TSJ. Third, for the same model, NVJ and TSJ vary widely across different accelerators, and adding GPUs can raise energy use for modest gains: eight H100s use 3.2x the energy of one to cut completion time by only 22%, at similar net value. More capable models and more hardware must therefore earn their extra energy with enough gains in output value. Finally, the work of experts who use AI is harder for AI to match than that of experts working alone. Measured against it, GLM 5.2's NVJ falls by 70% and its TSJ by 65%. Taken together, these findings justify economic return per joule as an objective for selecting professional tasks and improving language models and accelerators, and as a basis for examining AI's effect on the economy

0.25

Toward an Agentic AI Approach to Public Health Data Collection

cs.MA ·2026-10-22 ·Xueyi Zhang, Hongru Du, Ensheng Dong
benchmark
Abstract · 46 pages, 2 figures; Supplementary Information included

Disease surveillance increasingly integrates automated data collection, yet predefined sources and procedures require manual adaptation as reporting practices change. We explore agentic artificial intelligence (AI) as an approach to coordinating surveillance data collection across heterogeneous sources. Specifically, we built a multi-agent framework that integrates source discovery, source assessment, evidence structuring, event linking, and validation. Three retrospective applications illustrate its promise. The workflow recovered all 13 individual cases in a hantavirus outbreak, with links to 85 distinct publisher domains compared with 28 in the reference benchmark, reconstructed Canadian measles counts with a final total 0.33% below the national reference and higher cumulative counts on 79.8% of comparable days, and increased the share of days with recent Sierra Leone mpox reports from 28.2% to 65.6% by incorporating attributed media. These demonstrations also reveal limitations in record completeness and accuracy, while prospective gains in timeliness remain untested. We envision agentic AI as a complementary layer of public health data infrastructure, coordinating evidence-linked collection under surveillance goals and standards defined by public health professionals.

0.25

Operationalising Weaker Faithfulness: Recovery, Ambiguity and Computational Limits

cs.LG stat.MEstat.ML ·2026-10-22 ·Cillian Hourican, Rick Quax
Abstract

Variables can predict an outcome jointly while each is individually independent of it, causing structure learners to discard informative links. We investigate how to operationalise weaker faithfulness through two approaches: selective protection in an adjacency learner and explicit retention of alternative directed graph explanations. Matched experiments show that changes at both an early candidate-removal stage and later pruning can be necessary. In noisy XOR with a descendant (N = 10,000, 20 matched samples), combining reactivation and protection raises complete parent-pair recovery from 0/20 for either change alone to 20/20, while adding false edges. A confirmation study shows that the family estimator can preserve explanations lost by orienting the protected learner's selected skeleton: in one weaker signal XOR condition (N = 10,000) it retains the entire reference family in 48/50 samples, compared with 0/50 for the skeleton adapter. These gains are conditional. Protection can add only error, weaker effects produce false exclusions, and an ordinary faithful-chain comparison favours the PC learner. Complete nonempty family outputs decline at six/eight variables, and all 400 twelve/twenty-variable searches remain unfinished at the reporting cutoff. Neural controls show engineered retention, while the family estimator returns no accepted graph within its application budget. This operationalisation identifies two deletion stages that must be coordinated in the motivating example and shows when direct family search preserves explanations lost by selecting one skeleton. Finite evidence and computation remain substantial barriers to reliable general inference.

0.25

Topological Equivalence of Spectral and Density Cluster Hierarchies

cs.LG ·2026-10-22 ·Sajjad Hashemian
Abstract

We compare two canonical ways to recover hierarchical clustering from an underlying density $f$: the \emph{density cluster tree} given by connected components of superlevel sets $\{f\geλ\}$, and a \emph{spectral hierarchy} obtained by applying single linkage to diffusion distances induced by a Laplacian operator. We give conditions under which, at an appropriate diffusion time, the diffusion hierarchy is \emph{topologically equivalent} to the density cluster tree up to a monotone rescaling of heights. It further shows that a $k$-NN graph implementation inherits this equivalence from uniform approximation of diffusion distances, yielding a finite-sample reduction from spectral graph constructions to density-level-set topology.

0.25

Width and Depth Limits Commute During Training in Linear Residual Networks

cs.LG stat.ML ·2026-10-22 ·Ruijia Zhang, Soufiane Hayou
theoryhot direction
Abstract

Recent work established that the infinite-width and infinite-depth limits of residual networks commute at initialization when residual branches are scaled by $1/\sqrt{depth}$. In particular, the network output converges to the same limit whether width and depth tend to infinity sequentially or jointly. We extend this commutativity result to training in linear residual networks. Under the same $1/\sqrt{depth}$ residual-branch scaling (Depth-$μ$P) and the $μ$P width parameterization, we prove that, for any fixed number of gradient-descent steps with gradient clipping, the network output converges in $L^2$ along every joint width--depth growth path as both width and depth tend to infinity. Consequently, the infinite-width and infinite-depth limits commute throughout any finite training horizon. Without gradient clipping, the corresponding $L^2$ convergence does not hold in general, but we establish the weaker convergence in probability.

0.25

Identifiability from Stationary Covariance in Linear SDEs with Reciprocal Interactions and Latent Variables

cs.LG ·2026-10-22 ·Ana Leticia Garcez Vicente, Saber Salehkaleybar
Abstract

Continuous-time stochastic differential equations (SDEs) provide a natural framework for modeling interacting dynamical systems. A central causal question is whether causal effects among observed variables can be identified from observational data drawn from the stationary distribution, given the underlying causal graph. For linear SDEs, existing identifiability results are largely restricted to fully observed systems whose causal graphs contain no directed 2-cycles. We first establish new graphical conditions for the generic identifiability of causal effects in systems that may contain directed 2-cycles, thereby allowing for reciprocal interactions. Next, we address the previously unexplored setting of latent variables by developing an instrumental-variable framework for identifying a target direct causal effect under partial observability.

0.25

Deep ReLU Networks Can Fail the Best Approximation Property for Continuous Target Functions

cs.LG math.NAmath.OC ·2026-10-22 ·Sebastian Kassing
Abstract

We construct a continuous function on $[-1,1]^2$ for which the infimum of the $L^2$-approximation error over the class of ReLU networks with two hidden layers and architecture $(2,2,1,1)$ is not attained. This counterexample shows that deep ReLU networks do not satisfy the best approximation property in general, even for continuous target functions. Our findings contrast with existing results for shallow neural networks, where ReLU is one of the few activation functions that guarantee the existence of global minima in typical regression tasks. The proof is based on the fact that the breaklines for neurons in the second hidden layer can bend, which breaks the symmetry arguments used in the one-hidden-layer case.

0.25

Compatible Likelihoods for Flow Matching on Manifolds

cs.LG stat.ML ·2026-10-22 ·Lucas Ng, Georgios Batzolis, Mark Girolami
published
Abstract · 40th Conference on Neural Information Processing Systems (NeurIPS 2026)

On the simplest non-Euclidean manifold, the 1D flat torus, we exhibit a distribution where a recently proposed extension of Riemannian flow matching is globally optimized by a velocity field that pushes samples away from the data distribution. We trace this failure mode to a property we call compatibility; a likelihood-based flow matching objective is compatible if its population minimizer recovers the marginal velocity field induced by the data and coupling. We give a general condition for compatibility, apply it to existing methods, and bound the worst-case error in terms of a residual component vanishing under compatibility. Next, we note that existing compatible objectives put density on velocities that are infeasible on compact manifolds. We propose an objective built from an exponential family over endpoints, with the conditional velocity as sufficient statistic, which is both compatible and supported only on feasible velocities. This manifests as a reweighting of the flow matching loss which, on constant-curvature manifolds, treats errors in flow speed and direction differently. Empirically, our method recovers the correct velocity field on the test case, outperforms prior methods on high-dimensional product tori, and obtains substantial NLL improvements over prior work on two of four standard geospatial benchmarks on the sphere at moderate runtime overhead. On a large dataset of high-dimensional human poses, our method outperforms prior work according to sample-based metrics.

0.25

Rethinking Fisher Vectors for Second-order Multi-scale Deep Feature Aggregation in Scalable Scene Retrieval

cs.AI ·2026-10-22 ·Alexy Bhowmick, Sarat Saharia, Shyamanta M. Hazarika
Abstract · 8 pages, 4 figures

Aggregating CNN responses into a compact discriminative vector, given their lack of geometric invariance, is a challenge. To address this challenge, we propose a codebook-free method to encode second-order information of multi-scale CNN features within the Fisher Kernel framework. We use a scalable retrieval pipeline to show the efficacy of the proposed vector. The increased discriminative power is confirmed by experimental evaluation of our method on two difficult benchmarks (INRIA Holidays, UKBench). The significant improvement in retrieval accuracy over state-of-the-art approaches comes with only a slight dimension increase.

0.24

Action Learning for Scientific Simulations

cs.LG ·2026-10-22 ·Sarah Liaw, Yuanqi Du, Yilun Du
Abstract · 25 pages, 7 figures

Understanding and controlling how systems evolve is central to science and engineering, where many problems require reasoning about complete trajectories. Existing approaches typically learn local dynamics or generate trajectories, but objectives constructed from individual transitions can miss dependencies across time, while generative models do not necessarily provide a tractable objective for evaluating and optimizing arbitrary paths. We study $\textit{action learning}$, the problem of recovering a stochastic action from trajectory observations, yielding a common objective for reasoning about dynamical systems. We introduce $\textbf{LEAP}$ ($\textbf{L}$earned $\textbf{E}$nergy on $\textbf{A}$ction $\textbf{P}$aths), which anchors the unknown action to a tractable reference process and learns the remaining correction by comparing observed trajectories with reference trajectories. This construction yields a differentiable objective on path space that can capture both local dynamical structure and nonadditive dependencies across time. Once learned, the same action can be flexibly reused for path ranking and inference, mechanism steering, constrained optimization and sampling. Across stochastic physical systems, many-body orbital dynamics and molecular simulations, LEAP recovers underlying dynamics, captures dependencies across time and enables transfer across inference objectives and physical conditions. We show that $\textit{action learning}$ is a promising framework for turning trajectory observations into reusable objectives for scientific inference and control.

0.24

Directing to Diversify: Multi-Directional Search for Automatic Heuristic Design

cs.AI ·2026-10-22 ·Jianghan Zhu, Jianan Zhou, Fei Liu, Ni Zhang, Yaoxin Wu +2 more
Abstract · 46 pages, 9 figures, 19 tables. The main paper, including references, is 13 pages; appendices are included

Large language models (LLMs) have demonstrated strong potential to automate heuristic design for NP-hard combinatorial optimization (CO) tasks without human intervention. Recent methods predominantly rely on evolutionary computation (EC) frameworks, where heuristic diversification is mainly induced through EC-like operators or downstream selection, while the direction of heuristic transformation remains largely unharnessed. In this paper, we propose a multi-directional Perturbation-Rewrite Search (PRS) framework for automatic heuristic design that factorizes heuristic transformation through directional search primitives at the generation level. Specifically, PRS operates on heuristic ideas through three complementary directions: expansion toward richer and more expressive designs, compression toward simpler and less specialized designs, and steering for fine-grained refinement, together enabling broader and more sustained exploration of diverse regions in the heuristic design space. PRS further organizes candidate exploration within a breadth-first search-based process with adaptive operator control, which ensures thorough candidate utilization and sustained exploration, and maintains strong search dynamics while preventing premature convergence. Extensive empirical results across diverse CO tasks show that PRS consistently discovers high-quality heuristics, demonstrating the effectiveness and generality of explicitly directional exploration design.

0.24

ConsMSA: Semantic Distribution Consistency Learning for Multimodal Sentiment Analysis

cs.LG cs.MM ·2026-10-22 ·Pan Wang, Lipeng Ke, Huajun Ying, Pritish Mohapatra, Rohan Sarkar +3 more
published
Abstract · ICML 2026

Multimodal sentiment analysis (MSA) aims to predict human sentiments by integrating signals from different modalities such as text, video, and audio. However, raw multimodal sequences often suffer from semantic inconsistency--exhibiting redundancy or conflicts within and across modalities--which hinders robust understanding and increases computational cost. To this end, we introduce ConsMSA, which explicitly formalizes semantic distribution consistency across both $\textit{intra}$- and $\textit{inter}$-modality, providing a consistency-aware mechanism for robust and efficient multimodal sentiment prediction. Specifically, ConsMSA projects multimodal token features into a shared sentiment space to compute an Intra- and Inter-modality Consistency Score ($I^2CS$). By coupling this score with predictive relevance, we formulate consistency-aware importance signals that are utilized: (i) as a consistency regularizer to align latent distributions during training, (ii) to derive semantic-aware weights for adaptive multimodal token reweighting, and (iii) as a practical criterion to prune redundant or conflicting tokens. Extensive experiments on CMU-MOSI, CMU-MOSEI, and CH-SIMS demonstrate that ConsMSA achieves state-of-the-art performance while remaining robust under aggressive token compression--retaining only 10% of tokens yields comparable accuracy. These results establish semantic distribution consistency as a promising foundation for synergizing predictive robustness with computational efficiency.

0.24

The Bayesian Learning Rule Beyond KL Geometry

cs.LG ·2026-10-22 ·Sophia Sklaviadis, Wu Lin, Andre Martins, Mario Figueiredo, Thomas Moellenhoff +1 more
Abstract

The Bayesian learning rule (BLR) exploits the relation between Bayesian updating and the Kullback-Leibler divergence (KLD) to derive efficient additive updates for exponential family variational posteriors. The KL geometry is powerful but restrictive: combined with posterior families with parameter-dependent support it can lead to undefined objectives. We ask which non-KL geometries preserve the additive BLR computation. Previous works that use $f$-divergences suggest that this is only possible with the KLD. Here, we show that this is not true. We show that additive information-geometric optimization algorithms can be derived for the class of mixed-space Bregman or Fenchel-Young discrepancies. These discrepancies induce generalized maximum-entropy families, dual coordinates, and generalized-BLR canonical gradient updates, recovering the classical BLR. The generalized BLR leads to new algorithms that \textit{cannot} be derived from the KL-based BLR. For the Tsallis negentropy with quadratic statistics, the generalized BLR yields bounded-support and heavy-tailed $q$-Gaussian posteriors and an easy Newton-like implementation. Experiments up to CIFAR-100 illustrate controllable support and tail behavior and competitive performance.

0.24

CANNBot-CAKE: Collaborative Agentic Kernel Evolution for Ascend NPU

cs.AI ·2026-10-22 ·Deng Pan, Xue-Jian Gao, Junyi Fan, Bin Du, Jiasheng Li +6 more
Abstract · 23 pages, 10 figures

Efficient custom operators are essential for modern deep learning workloads, yet developing high-performance kernels for Ascend NPUs remains slow, expert-intensive, and difficult to automate. Although AI agents now approach expert performance on GPU kernel benchmarks, current models remain less capable of generating AscendC kernels for NPUs, potentially reflecting the relative scarcity of relevant training data. We thus present CAKE, a CANN-native system for deployable custom operators that combines adaptive multi-DSL routing, separate Generation and Evolution modes, and reviewed knowledge reuse through CAKE Book and a Self-Evolving Agent. On 53 CANN-Bench v1.1 operators on Ascend 950PR, CAKE's aggregated results achieve $100\%$ case pass rate on the standard set (1,060/1,060) and $98.28\%$ on the hidden set (4,167/4,240), with an overall mean score of 78.63 out of 100 and an average performance speedup of $1.334\times$ over the baseline. Evolution then pushes correct operators further: EngramGateFusion cuts aggregate device time by $46.6\%$ over 20 rounds and roughly halves the mean roofline gap. These results offer a practical path toward closing Ascend's AI-for-kernel capability gap.

0.24

When News Gets Old: Portfolio Allocation with Reinforcement Learning and Large Language Models

cs.LG q-fin.PM ·2026-10-22 ·Bolin Song, Surui Xiang
Abstract

Financial news can influence investment decisions long after publication. Timely news may reach an investor before prices fully adjust, whereas older news may already be reflected in prices. We examine two ways to handle aging signals: retain them and give a portfolio policy timing information, or filter them out before portfolio selection. A fixed large language model (LLM) extracts signals from public SEC disclosures. We compare these treatments in reinforcement learning (RL) policies and LLMs that choose portfolios directly. We evaluate returns after trading costs in historical tests over 2023-2025. RL choices respond to timing information in a controlled test, but historical returns show no consistent benefit from adding it. For the LLM, filtering old news produced the highest average net return, reaching 18.43% per year when news delivery was delayed by four weeks. This exceeded the returns of all RL strategies that used news.

0.24

Agri-SCHOLAR: Simulation-Grounded In-Season Rolling Advisories for Agricultural Crops Using LLM-Bounded Optimization

cs.AI cs.MA ·2026-10-22 ·Geetha Charan, Vedant Balasubramaniam, Manojkumar Patil, V. Priyanka, Kodur Sai Vinay Sathvik +3 more
Abstract

Existing agricultural advisories to farmers suffer from several limitations: (a) remain largely expert-driven, (b) do not adequately account for in-season weather, and (c) do not scale to individual farms. In this paper, our objective is to close these gaps by developing Agri-SCHOLAR (Simulation-Grounded Crop Harvest Optimization with LLM-Guided In-Season Advisory Rollouts). Agri-SCHOLAR is a rolling, in-season advisory framework that issues, at intervals of every few days or every week, a personalized advisory using the following information: soil and crop properties, observed weather to date, short-range forecast of weather, and a season extrapolation of weather based on historical analogs. The novelty and depth of Agri-SCHOLAR lies in each advisory being produced by a closed-loop system which couples multiple LLM agents with a powerful crop simulator (such as APSIM in our case). The simulator crucially serves as an evaluation oracle. In Agri-SCHOLAR, the LLM agents propose agronomically sound and feasible bounds on actionable parameters and an optimizer searches within those bounds against a multi-objective reward spanning yield, resource utilization, and economic feasibility. We conduct a detailed 11-year analysis of the maize crop in two selected locations of South Asia and show that Agri-SCHOLAR consistently outperforms the region's standard packages of practice and satisfactorily addresses the limitations of existing agricultural advisories.

0.24

Federated Graph Retrieval-Augmented Generation

cs.LG ·2026-10-22 ·Xiaobao Wang, Renda Han, Zhiliang Xia, Meng Liu, Renxiang Guan +1 more
Abstract

Graph Retrieval-Augmented Generation (GraphRAG) leverages centralized knowledge graphs to enhance large language models with structured knowledge retrieval. However, real-world knowledge is inherently distributed across multiple organizations rather than being centrally accessible. Privacy constraints further prohibit sharing entire local knowledge graphs, preventing GraphRAG from effectively retrieving and utilizing distributed knowledge. To this end, we propose Federated GraphRAG (FedGRAG), a unified framework for distributed graph retrieval and knowledge-enhanced generation over multiple clients with knowledge graphs. Specifically, we address two fundamental challenges in federated GraphRAG. First, the absence of a global knowledge graph makes unified retrieval impossible. To address this, we develop a Federated Knowledge Routing mechanism that efficiently identifies relevant clients through lightweight knowledge indices without exposing local graph structures. Second, local graph memory cannot be shared directly because it contains sensitive entities and knowledge information. We therefore propose a controlled-disclosure evidence construction method. It extracts retrieved evidence subgraphs and converts them into an evidence abstract, enabling evidence conversion and reconstruction on the server. Extensive experiments demonstrate that Federated GraphRAG consistently outperforms existing baselines in both retrieval and downstream generation tasks, establishing a new paradigm for deploying GraphRAG in federated environments.

0.24

Algorithmic drift in generative adaptive learning

cs.AI ·2026-10-22 ·Firuz Kamalov, Hana Sulieman, Ahmed El Sayed, Sivaraj R., Nitesh Sughnani
benchmark
Abstract

Generative artificial intelligence promises to democratize personalized tutoring, yet the longitudinal stability of adaptive systems remains under-theorized. Current ethical discourse focuses on static bias. This paper argues that a distinct threat lies in algorithmic drift: a feedback loop in which minor, demographically asymmetric observation errors compound over time into systemic inequities. We make three contributions. First, we formalize algorithmic drift as a coupled learner--tutor dynamical system. Second, we corroborate this theory with an agent-based Monte Carlo simulation in which either a transparent online tracker or an LSTM deep-knowledge-tracing network drives an IRT-based curriculum policy. Third, we introduce the temporal fairness stabilizer (TFS), a targeted intervention that periodically re-probes learner capability, and benchmark it against standard exploration baselines with explicit group-fairness metrics and replication-level inference. Within the modeled environment, TFS closes the achievement gap where generic exploration does not, and we characterize the conditions under which it succeeds. We separate what is established analytically, what is demonstrated in simulation, and what remains to be validated with real learners. Our results reframe algorithmic fairness in education as a temporal property of a closed loop rather than a deployment-time checkbox.

0.24

Evaluating the Applicability of LLM Inference Optimizations on Apple Silicon

cs.LG cs.DC ·2026-10-22 ·Kyrylo Yemets, Andrii Yaroshevych, Victor Muryn, Ostap Khomenko, Pavlo Kryven +3 more
publishedbenchmark
Abstract · Paper was accepted at the NeurIPS 2026 Workshop on On-Device Intelligence (ODI) and the NeurIPS 2026 Workshop on SLMs for Agentic Systems (S…

Apple Silicon is among the most widely used platforms for local LLM inference, yet nearly all inference optimizations are designed and validated on server-class NVIDIA GPUs under batched serving workloads. Which optimizations retain their benefits when moved to Apple Silicon, and what determines their applicability? We evaluate four optimization families: KV-cache pruning, speculative decoding, ported GPU kernels, and post-training quantization. Experiments span four Apple machines under a common single-request decode workload, with matched RTX 4090 comparisons where possible. KV-cache pruning can reach up to 1.75 times end-to-end throughput on Apple but can reduce throughput on desktop NVIDIA GPUs. Under the common benchmark, all evaluated speculative decoding methods reduce throughput on Apple, despite reported 2-6 times gains on server GPUs. CUDA kernel designs transfer only partially: public Metal lacks asynchronous copy, and batch-one decode makes several serving-oriented mechanisms inapplicable, although transferable components yield up to 1.20 times decode throughput. Quantization gains vary by machine. Across four families, a reported speedup is not a portable property of a technique alone but a joint function of technique, runtime, hardware contract, and workload regime.

0.24

Evaluating Context Sensitivity of Collaboration Behavior in Multi Party Simulation Frameworks

cs.MA cs.AI ·2026-10-22 ·Anvesh Rao Vijjini, Sheshera Mysore, Pegah Jandaghi, Ashish Sharma, Snigdha Chaturvedi +1 more
Abstract

Group Social Behavior is increasingly used as a lens to study Large language model (LLM) simulations to aid model training and evaluation. However, little prior work has examined how contextual factors such as team size and goal type manifest in long running simulations. To overcome this gap, we conduct a controlled, full-factorial study examining how contextual factors affect collaboration behavior across multi-party simulation frameworks. Using well established social science taxonomies, we quantify context sensitivity of collaboration behavior in simulated teams. Our analyses reveal significant effects of contextual factors such as simulators consistently favor smooth collaboration over disagreement, negotiation pressure, and coordination breakdown. These findings provide the first systematic evidence of how contextual factors influence emergent collaboration in LLM based multi-party simulation frameworks, offering both guidance and empirical results for the use of such simulations.

0.24

Can AI Co-Design Structures? An AI Agent for Early-Stage Structural Design

cs.AI ·2026-10-22 ·Ricardo Maia Avelino, Mennatallah El-Assady, Tom Van Mele, Philippe Block
published
Abstract · Paper accepted to the International Association for Shell and Spatial Structures (IASS) Annual Symposium 2026, in Torino, Italy

Recent advances in large language models and agentic workflows offer a new paradigm to early-stage structural design translating design intent into structural models. Although text-to-3D approaches have demonstrated that natural language can be used to produce visual, mesh-based geometric models, structural design requires domain-specific geometric arrangements and semantically rich data types. This paper introduces a proof-of-concept workflow for prompt-to-structure generation using an AI agent connected to a COMPAS-based structural modelling API through a Model Context Protocol (MCP) server. Through controlled modelling operations, the agent creates and modifies a graph-based structural data model composed of explicit building elements. The resulting artifacts can be inspected, modified, serialised, and further connected to downstream AEC workflows. The approach is demonstrated through selected prompt-to-structure examples and by progressive and editable design sequences that resemble typical iterative structural design workflows. The results are preliminary, but demonstrate the possibility of combining natural language, agentic workflows, and domain-specific structural data structures as a first step toward AI-assisted structural co-design.

0.24

Primal-Dual Representation Learning for Low-Rank Constrained MDPs

cs.LG ·2026-10-22 ·Chenhao Zhou, Chao Zhang, Wensong Bai, Hanbin Zhao, Hui Qian
top venuetheory
Abstract · Accepted at NeurIPS 2026

Representation learning has substantially improved the efficiency of unconstrained continuous-state Markov Decision Processes (MDPs), but its extension to Constrained MDPs (CMDPs) is difficult because the transition representation must be learned while constraint violation is controlled. In this paper, we study representation learning for low-rank CMDPs with continuous state spaces, and propose REP-PD-TEG for the soft cumulative-violation setting. In the feature learning period, the algorithm interacts with the environment using a decaying-$ε$-greedy mixture execution policy, and uses the collected samples to learn transition features by maximum likelihood. Based on the learned features, it constructs uncertainty bonuses and plans with an optimistic Lagrangian objective to obtain the return policy. We prove regret and soft-violation guarantees for the return policies, and quantify the influence of the decaying-$ε$-greedy exploration. In particular, REP-PD-TEG achieves the optimal $\widetilde O(\sqrt K)$ return policy bound, and gives a $\widetilde O(K^{3/4})$ overall sample bound when exploratory actions are charged. For the stricter hard cumulative violation, where feasibility must be verified episode by episode, we propose REP-PD-H-TEG. It plans under a bonus-corrected lower-confidence utility certificate with the learned representation, and performs a stabilized dual search to obtain the return policy, yielding regret and hard-violation bounds. Together, these results give the first theoretically guaranteed representation-learning method for low-rank continuous-state CMDPs with both soft cumulative guarantees and certified hard-violation control.

0.24

Can Loan Processors and Underwriters Spot AI-Edited Statements? An Empirical Study

cs.AI cs.CRcs.HCcs.SE ·2026-10-22 ·Simiao Ren, Xingyu Shen, Tommy Duong, Ankit Raj
Abstract · 6 pages, 2 figures

AI-assisted editing can produce loan documents that preserve familiar issuer layouts while changing the applicant's identity. We investigated whether experienced underwriters and loan processors identify such forgeries in first-pass, document-only review. Three practitioners recruited through Upwork each received one applicant package drawn from 134 forged PDFs derived from 47 genuine financial documents supplied with full consent. The edits preserved amounts, dates and balances; identifier inconsistencies introduced by the pipeline were knowingly retained before distribution. Reviewers were not told that files had been altered and returned 132 file-level judgements. Checking their explanations against the originals yielded 14 edit-supported concerns (10.6%): 12 Suspicious judgements and two Insufficient information judgements. All supported concerns involved identifier disagreements within or across documents, and no reported visual cue identified an actual edit. Thus, approximately one in ten judgements contained evidence of the forgery, despite known inconsistencies remaining in the files. These observations motivate evaluating consistency checks and independent source verification alongside human review. With three reviewer-package pairs and no genuine-file controls, the findings describe this exploratory setting rather than population-wide detection accuracy.

0.24

Benchmarks Are Surprisingly Compaction-Friendly

cs.LG cs.AI ·2026-10-22 ·Jeanne Salle, Hardik Bhatnagar, Matthias Bethge, Sahar Abdelnabi, Maksym Andriushchenko
position
Abstract · 25 pages, 11 figures

Long horizon agentic tasks are now pushing LLM agents past their context window. To continue the task, the agent scaffold has to free up the context window, a process known as context compaction. Compaction is generally lossy by design: agents can silently lose task constraints, instructions, or their reasoning behind earlier decisions. Although compaction is a standard component of coding agent scaffolds, it has not been studied systematically with frontier models on realistic, long horizon agentic tasks. Previous work evaluate smaller open weight models on toy or synthetic trajectories. We fix the agent scaffold (OpenCode), implement five compaction strategies inside it, and evaluate three frontier models on three long horizon benchmarks (PaperBench, ProgramBench, PostTrainBench). For the two 1M context models, the compaction threshold does not measurably affect the task performance, whereas GPT-5.3-Codex, the model with the smallest context window, performs worse at lower thresholds on PaperBench and, less clearly, on PostTrainBench. No compaction method outperforms default self-summarization, and methods that discard more content cost some models up to 20 points while leaving other models the same. Compaction can, however, silently remove task constraints: on ProgramBench, the first compaction summary already drops about one third of the task's anticheating rules on average, and this rises monotonically with the number of compactions. On real user-agent sessions from SWE-chat, instructions about how the agent should work and what it should not do are dropped around 6 times more than specifications of what to build. We argue that current benchmarks are compaction-friendly: i.e., the objective is fixed and the agent's progress stays on disk, so most of what compaction discards can be recovered, except constraints that don't exist anywhere else.

0.23

Reproducible Evaluation as the Substrate for Safe Prompt Compression in Production Agents

cs.AI ·2026-10-22 ·Prabal Vashisht, Zheng Hu, Ying Yang
publishedbenchmark
Abstract · 8 pages (4 pages main text plus references and appendices), 1 table. Accepted at the NeurIPS 2026 Workshop on Agents in the Wild

Evaluating conversational agents that act on live, mutable enterprise data is hard: replayed sessions drift as backend state changes, and a benchmark built today can silently go stale as the data it describes moves on. We present an offline evaluation harness for a production conversational agent that stays reproducible despite the live data it operates on changing, and that can be regenerated on demand once it goes stale. We then use this harness as the fitness function for GEPA-based prompt optimization under a correctness gate that guards against reward hacking. Applied to a production merchant-facing agent, correctness-gated compression shrinks the agent's full prompt surface by 3.5-7.6% without regressing held-out task correctness, consistently across three independent subject-model families, and for one, correctness measurably improves.

0.23

Training Small Language Models for Transferable Tool Selection

cs.LG cs.AI ·2026-10-22 ·Zhaoyi Zhou, Akshayaa Magesh, Jeongyeol Kwon, Artem Kuriksha, Daniel Jiang +2 more
Abstract · 19 pages, 5 figures, 5 tables

As LLM agents operate over corpora of more than ten thousand tools, selecting the right tool for a complex multi-step user instruction becomes a central bottleneck. Tool descriptions cannot all fit in context, and relevance-based retrievers only rank tools without knowing which are actually needed. Recent work instead leverages proprietary commercial LLMs like Claude or ChatGPT to drive a retriever, decomposing the instruction and selecting over retrieved candidates. We show that a small, open model can be trained via reinforcement learning to perform the same functionality. On ToolBench, a Qwen3-8B model post-trained with our recipe outperforms Llama-4-Maverick (402B). In addition, the trained model transfers directly to different retrievers and generalizes to new tool corpora in different domains.

0.23

Piano2Pop: Expanding Piano Demos into Pop Productions Without Aligned Supervision

cs.LG cs.SD ·2026-10-22 ·Yun-Chen Cheng, Chih-Pin Tan, Tzu-Hung Huang, Yi-Hsuan Yang
Abstract · Submitted to ICASSP2027

We introduce Piano2Pop, a music-expansion task that transforms piano MIDI into instrumental pop productions. Enabled by recent advances in music generation and polyphonic music transcription, our pseudo-supervised approach conditions a pretrained flow-matching model on piano rolls without aligned piano-pop supervision. We train on synthetic piano-instrumental pairs derived from pop recordings, comparing pair-free transcription, including function-selective (FST) and register-density (RDA) augmentations, with a pair-supervised piano-cover generator as pseudo-demo sources. On 476 testing real-world piano-cover/pop pairs, the pair-supervised approach achieves the best Frechet Audio Distance, Accompaniment Prompt Adherence, and spectral shape on real-piano inputs, while FST mitigates piano-heavy outputs in pair-free training and remains a promising alternative. All fine-tuned systems significantly outperform the zero-shot baselines, which suggests a role for task-specific modeling.

0.23

Decode Late, Multiply Once: Rethinking the Dot Product for Quantized Weights

cs.LG cs.AR ·2026-10-22 ·Jonathan Tapson
Abstract

Large neural networks are increasingly deployed with 1-4 bit weights and 16- or 32-bit activations, yet their dot products still use one multiplier per element. We show that a length-n dot product with k-bit weight codes can be computed with n+2^k additions and at most 2^k decodes and 2^k multiplications. Each weight code is used directly as an address offset that selects one of 2^k accumulators; each activation is added into the selected accumulator; and the accumulated sums are decoded and multiplied only once, at the end, in a fused decode-and-multiply step. The decoded weight vector is never formed. The refactoring is exact for integer arithmetic, works with integer or floating-point operands and with arbitrary, including nonlinear, decoders, and extends to group-wise scales and to matrix and tensor products. The multiplier count falls by n/2^k, while total arithmetic energy improves by a smaller, bounded factor that we quantify with a cost model. We describe a scatter-accumulate processing element (SA-PE) that replaces the MAC unit in a weight-stationary systolic array, with a load/stream/reduce/drain schedule in which a small, time-shareable multiplier runs only 2^k times per tile. Because decoded weights are not stored, local weight memory shrinks by up to w_d/k for w_d-bit decoded weights. An RTL implementation synthesized for FPGA and compared with a matched MAC array shows that an unshared SA-PE is larger than a MAC PE, and that the saving appears when the reduction multiplier is shared across PEs. Software models confirm exact integer equivalence and floating-point error no worse than the conventional pipeline. We do not claim the underlying identity is new, and we discuss related work on weight sharing and lookup-based kernels.

0.23

Sharp Bounds on the Winner's Bias under Independent and Nonnegatively Correlated Noise

cs.LG math.ST ·2026-10-22 ·Ivan Viakhirev, Kirill Borodin, Grach Mkrtchian
Abstract

Reporting the best of $M$ noisy, unbiased scores overstates the winner's value by up to $c\sqrt{V}$, where $V$ is the total noise variance and the sharp constant $c$ depends on how the noise is dependent. It was known only at the two extremes, arbitrary dependence ($\sqrt{(M-1)/M}$) and i.i.d. noise. We determine it for two classes in between: for independent, not identically distributed noise it is $\sqrt{(M-1)/(2M)}$, and for pairwise nonnegatively correlated noise it is $(M-1)/M$; for uncorrelated noise this answers the maximum version of an open question about the expected range. Explicit finite constructions attain both. As a corollary, the bias of the cross-validation estimator is bounded sharply within each class. All theorems are formally verified in Lean 4. Across 448 language-model and classifier selection pipelines, designs with per-candidate items realize a median 0.75 of the independent-class ceiling. At a fixed item budget, scoring all candidates on shared items is markedly less optimistic than giving each candidate its own items.

0.23

Certified Execution Harness for Agentic AI: Safe and Least-Intrusive Petri-Net Supervisory Control Between Reasoning and Action

cs.AI eess.SY ·2026-10-22 ·Chao Gu, Quanyan Zhu
Abstract · 17 pages, 9 figures, 2 tables. Extended version of a paper submitted for publication

This paper studies the problem of safe tool execution for large language model (LLM) agents under prompt injection. Using Petri nets, we formalize the execution harness between the LLM and its tools as a supervisory control problem, in which marks represent capabilities, approvals and budgets. Since a transition fires only if the marks it needs are present, an injected request whose preconditions are unmet or already spent is refused regardless of its wording. Treating tool outcomes as uncontrollable, we enforce safety by the conventional maximally permissive supervisor, and this safety transfers to the running system under four assumptions. Moreover, to capture an agent that may propose again after a refusal, we introduce the notion of a replanning closed loop, in which each retry costs an inference call. We show that, in this loop, the conventional supervisor can be arbitrarily more intrusive than a stricter safe one. Furthermore, for a given proposer model, the maximally permissive least-intrusive supervisor is characterized as a value-threshold refinement of the conventional one. Under precise conditions, this refinement orders two concurrent actions by at most one monitor place. On an autonomous release agent, we find by synthesis that safe deployment requires a step that the supervisor can refuse after a rollout starts.

0.23

Reward-Weighted Preference Optimization for Selective Combinatorial Optimization Problems

cs.AI ·2026-10-22 ·Haohui Zhang, Xinyu Hu, Wouter van Heeswijk, Neil Yorke-Smith, Martijn Mes
Abstract · 37 pages, 12 figures

Reinforcement learning has emerged as an effective approach for combinatorial optimization, but most existing methods focus on problems such as the Travelling Salesperson Problem (TSP) in which all elements must be served. Selective combinatorial optimization problems (SCOPs), including the Orienteering Problem (OP), Prize-Collecting TSP (PCTSP), Stochastic PCTSP and Knapsack Problem (KP), additionally require selecting which elements to include. This selection creates sampled solutions with substantially different objective values, yet standard preference optimization reduces such comparisons to binary rankings and discards the magnitude of these differences. We introduce Reward-Weighted Preference Optimization (RWPO), a pairwise training objective that incorporates reward-gap information into preference learning. For candidate trajectories sampled from the same instance, RWPO weights each preference loss according to the normalized difference in their rewards, assigning greater training emphasis to comparisons that represent larger improvements in solution quality. We further combine RWPO with scale-specific fine-tuning and reward- and likelihood-weighted tilted sampling to adapt trained policies to larger and shifted instance distributions without relying on local-search post-processing. Experiments on the OP, PCTSP, Stochastic PCTSP, and KP show that RWPO consistently improves over unweighted preference optimization and policy-gradient baselines across a wide range of instance sizes. On the OP, the resulting solver exceeds the mean reward of the Compass metaheuristic at scales from 100 to 2,000 nodes. The results demonstrate that preserving reward-gap information can substantially improve preference-based learning for selective combinatorial optimization and support generalization to large-scale instances.

0.22

Is High Local Accuracy Enough? Geometric Consensus with Support-Adaptive Statistical Personalization for Federated Learning

cs.LG ·2026-10-22 ·Junyu Lin, Bosong Qi, Pengcheng Shi, Jiayuan Li
code/data
Abstract · 27 pages, including 9 pages of supplementary material. Code: https://github.com/JL181/FedGRC-Benchmark

Real-world deployment of personalized federated learning (PFL) requires adapting to heterogeneous clients while preserving recognition performance beyond local distributions. However, high local accuracy can conceal failures on locally rare or absent classes. Effective personalization should therefore strengthen global recognition while raising the personalization floor. We propose an evaluation protocol that complements local accuracy with global, rare-class, and unseen-class evaluations to reveal generalization obscured by local scores, and assesses personalization floors among the worst-performing clients to expose weaknesses hidden by aggregate performance. To meet these deployment challenges, we propose FedGRC, a unified PFL framework integrating generalization and personalization through (i) Consensus-Guided Representation Learning (CGRL), which learns discriminative representations through shared directional and magnitude guidance; (ii) Multilayer Metric Synthesis (MMS), which constructs a shared covariance metric from these multilayer representations for statistical classification; and (iii) Support-Adaptive Shrinkage Personalization (SASP), which adapts class means within this metric by shrinking local estimates toward shared means according to local sample support, enabling adaptation while retaining shared knowledge. Extensive experiments, including benchmarking against 21 methods across 5 heterogeneity settings on 3 datasets and comprehensive ablation studies, validate the effectiveness of FedGRC. On CIFAR-10 and CIFAR-100, FedGRC surpasses state-of-the-art (SOTA) baselines in global recognition and personalization floors by at least 2.17% and 2.13%, respectively. On the real-world heterogeneous FedRS dataset, it outperforms the respective best baselines by 2.15%, 3.30%, and 4.61% in local adaptation, global recognition, and the personalization floor.

0.22

Autoregressive Multimodal Anomaly Detection with Large Language Models for Microservice Systems

cs.LG cs.AIcs.SE ·2026-10-22 ·Yixiang Tang, Fagui Liu, Guoxiang Zhong, Xi Yao, C. L. Philip Chen
Abstract · 14 pages, 8 figures, 7 tables

Microservice systems have become a dominant architecture for large-scale cloud applications, enabling rapid deployment and elastic scaling. However, ensuring the reliability and availability of microservice systems critically relies on effective anomaly detection, as failures and performance anomalies can quickly propagate across services and severely impact end-to-end system behavior. Existing multimodal anomaly detection approaches for microservice systems often struggle with two fundamental limitations: (i) the neglect of modality-specific characteristics in favor of cross-modality fusion, and (ii) the "identity-shortcut" problem that affects the feature reconstruction process. To address these issues, we propose LLMMAD, an autoregressive multimodal anomaly detection framework for microservice systems based on Large Language Models (LLMs). Initially, to reconcile cross-modality fusion with modality-specific preservation, LLMMAD embeds and recombines multimodal data into a unified sequence for LLM processing to model inter-modality relationships, while employing modality-wise reconstruction heads to retain modality-specific information. Subsequently, to eliminate the identity-shortcut, we implement a prediction-driven autoregressive objective, under which the model predicts future states rather than copies current inputs. Finally, we conduct extensive experiments on three real-world datasets. The results demonstrate that LLMMAD consistently outperforms state-of-the-art methods, achieving F1-score improvements of 3.1%, 1.4%, and 0.9%, respectively, thereby providing a more reliable solution for microservice monitoring.

0.22

Local Discrimination of Binary Classifiers via Localized AUC Functional

cs.LG stat.ML ·2026-10-22 ·Wenye Ren, Anna Győrffy-Kerekes, Tancredi Dorsi, Olga Demler
Abstract

Risk prediction models are frequently evaluated with AUC, which means that during evaluation we are blind to performance heterogeneity, which could lead to bad decision-making in clinics. In this work, we aim to change that by identifying patient cohorts for whom the risk prediction model might not be an accurate predictor. To separate these cohorts, we use a version of the weighted AUC, which we call Localized AUC or LAUC. Using this metric, we build a method that we coin LOCATOR. This method is able to identify cohorts of patients for whom the risk prediction model is inaccurate. We test this method both in real-world data and in simulations. We also build a statistical background for the use of Localized AUC for our purposes.

0.22

A Mathematical Theory of the Jacobian Lens

cs.LG cs.AIq-bio.NC ·2026-10-22 ·Mitchell Ostrow, Daoyuan Qian, Toni Liu, Ila Fiete
Abstract · 27 pages, 19 figures

Recent work from Anthropic introduced a new interpretability tool, called the Jacobian lens (J-lens, Gurnee et al., 2026). This method identified linear directions in the hidden states of frontier language models that had properties reminiscent of working memory-like representations. However, it is unclear what mechanistic properties of such features make them detectable to the J-lens. To this end, we introduce a mathematical theory of the J-lens, and show that it preferentially identifies coherent features -- directions that act like leading, spatially coherent Lyapunov modes. Small perturbations along these directions amplify across layers, remain roughly constant across token positions, and are consistent across different contexts. While we are agnostic to whether these properties are necessarily associated with a ``Global Workspace'', the analytically derived preferences of the J-lens nevertheless explain prior empirical applied to both model classes. We validate our analysis on a Bayes filter and a toy transformer, each optimally predicting next tokens emitted from a Hidden Markov Model (HMM). We show how modulating task parameters that change the coherence of features in the Bayes filter correspondingly and predictively modulates the response of the J-lens . We also introduce of a family of J-lenses built to emphasize different temporal scales of information processing in LLMs, which we call $ω$-J-lenses.

0.22

Metric Conformal Prediction via the Conditional Expected Local Radius

cs.LG stat.ML ·2026-10-22 ·Rui Qiu, Fang Yao, Zhou Yu
Abstract

We introduce the conditional expected local radius (CELR) and develop a new conformal prediction method for non-Euclidean responses in general metric spaces. CELR measures local conditional probability concentration, allowing prediction sets to accommodate heterogeneity and multimodality without requiring densities or a vector space structure. Its conditional distribution yields a nonconformity score that is comparable across covariate values and avoids costly transport comparisons between distance profiles. We further adapt a metric distributional random forest that handles multivariate covariates and nonlinear interactions, while a single forest fit supplies all score components, reducing computation by eliminating repeated local fitting. We establish uniform consistency of the estimators and asymptotic conditional validity under suitable regularity conditions, together with finite-sample marginal coverage under exchangeability. Numerical studies demonstrate effective uncertainty quantification across different metric spaces and substantial computational savings over the conditional profile score method.

0.22

Direct Inference on Admissible Sets (Diona)

cs.LG math.OCeess.SY ·2026-10-22 ·Lorenzo Sabug, Eric Kerrigan
benchmark
Abstract · 27 pages, 4 figures

We propose a new regression method named Direct Inference on Admissible Sets (Diona) that directly uses samples to estimate the state and its derivatives at a query point, along with their admissible bounds, without long global training. By casting the values and derivatives at the query and neighbouring samples as latent variables in a constrained optimisation problem, the burden of long training loops becomes one of on-demand local solves. We establish Diona's theoretical properties including finite-sample coverage and, in the noiseless case, convergence to the true state and derivatives under dense sampling. In a nonlinear regression benchmark on six functions of three to eight dimensions, Diona was more accurate than a Gaussian process (GP) on noiseless data especially with denser data sets. Furthermore, as with GP, it produced uncertainty bounds that contained the true value at more than 95% of query points in four functions. Diona's time to a first prediction after a data set update was three to four orders of magnitude shorter than that of the GP, and two to three orders shorter than that of a neural network, which was more accurate but offers no uncertainty quantification. This low latency is paid for by on-demand predictions, so this latency advantage applies when relatively few predictions are needed between data updates.

0.22

Simulated Control Evals

cs.AI ·2026-10-22 ·Jeffrey Olmo, David Zhang, Tyler Tracy
Abstract

Control evaluations measure whether safeguards can prevent an AI model from causing undesired outcomes even when it is intentionally trying to do so. These evaluations typically use purpose-built control settings and extrapolate their results to a model's deployment. However, the effectiveness of safeguards in deployment can depend on features of environments and tasks that control settings do not represent. To broaden where control evaluations can be run, we introduce simulated control evaluations (SCEs), which use a language model to simulate tool responses in place of an environment. We evaluate SCEs on LinuxArena, where execution provides ground-truth outcomes. We compare SCE prediction errors with those from a separate control setting, BashArena, and with variation in control metrics within LinuxArena. For independently sampled evaluation and deployment task sets, SCEs predict safety with a root mean squared error of 3.6 percentage points across five monitors at a 5% audit budget, compared with 10.1 for BashArena. SCEs closely reproduce distributions of attack strategies, attacker reasoning, and monitor reasoning, with over tenfold lower KL divergence from execution than BashArena. SCEs explain 78-85% of the between-environment variance in attack success, monitor scores, and safety. This accuracy is sample efficient: SCEs predict safety with less error than BashArena using fewer than 5% as many task-pair draws as the deployment task set. Strategies discovered through simulation also improve monitor performance on held-out LinuxArena attacks, outperforming BashArena and approximately matching improvement from executed strategies. These results suggest that deployment-specific SCEs can be more predictive than control evaluations executed in separate control settings.

0.22

Anticipating Social Consequences with Distributional World Models

cs.AI ·2026-10-22 ·Viet-Thanh Pham, Thuy-Trang Vu, Zhuang Li, Lizhen Qu, Gholamreza Haffari
Abstract · 29 pages, 8 figures

Social world models predict how interactions develop after an action, helping agents negotiate, cooperate and mediate. In these situations, one action can lead to several plausible futures, hence predicting only one future can overlook consequential alternatives, while exploring them through repeated simulations increases inference cost. We propose DSWM, a Distributional Social World Model that predicts alternative responses, their probabilities and immediate and final outcome values in one output per candidate action. We train a 4B model on repeated simulations using supervised fine-tuning and GRPO. At inference time, the actor uses these possible consequences to choose an action. Across three seen and two unseen environments, with interaction backbones absent from training collection, DSWM achieves the best average performance on 13 of 15 metrics. Ablations show that retaining multiple futures and their predicted probabilities improves outcomes in all five environments.

0.21

Different Tasks, Different Timesteps: Temporal Task-Aware Quantization for Diffusion Language Models

cs.AI ·2026-10-22 ·Ofek Avivi, Eliron Rahimi, Avi Mendelson, Chaim Baskin
Abstract · preprint

Diffusion Language Models (DLMs) introduce unique quantization challenges due to their iterative denoising process. In this work, we reveal that quantization sensitivity is both taskand time-dependent, with different downstream tasks exhibiting distinct precision requirements across denoising steps. We further show that these patterns can be predicted without groundtruth labels or task-specific evaluation metrics. Building on these findings, we introduce T 2Quant, a task and time-aware mixedprecision quantization framework that uses a label-free sensitivity signal to optimally allocate activation precision across denoising steps under a fixed bit-width budget. Extensive experiments across three DLMs and six benchmarks demonstrate that T 2Quant consistently improves the average accuracy of existing quantization methods, achieving gains of up to 6.7 percentage points under the same budget.

0.21

Direct Pass@$k$ Optimization for Thought Pruning: Balancing Quality and Diversity in Test-Time Reasoning

cs.AI cs.CL ·2026-10-22 ·Mahdi Salmani, Adel Javanmard, Devansh Gupta, Meisam Razaviyayn
Abstract

Test-time scaling has emerged as a key inference strategy for reasoning models, where a model generates a large number of candidate solutions in the hope of finding at least one correct solution. Running all candidate solutions to completion can be computationally demanding, so recent work prunes the candidate pool after generating only a short prefix, committing further computation only to the surviving candidates. Existing pruning strategies divide broadly into two categories: Quality-based methods assign a score to each partial solution and discard those predicted to ultimately lead to an incorrect final solution, while diversity-based methods select partial solutions that follow different reasoning trajectories and are therefore likely to cover different final solutions. The tension between these two objectives is an instance of exploration versus exploitation, yet little to no prior work on pruning exposes this trade-off directly, let alone optimizes it. In this paper, we introduce a pruning procedure that addresses this trade-off directly by optimizing pass@$k$ over the selected solution pool. In particular, it chooses the pool to maximize the probability that at least one of its partial solutions, when continued to completion, yields a correct answer. Because this objective simultaneously rewards high-quality partial solutions and penalizes redundancy among them, quality and diversity are naturally balanced by the criterion itself rather than through separate ad-hoc design choices. On eight open-weight reasoning models, our method outperforms existing quality-based and diversity-based pruning strategies. The gains are particularly pronounced when the pruning procedure keeps a moderate number of partial solutions for continued generation, highlighting the value of directly balancing quality and diversity during pruning.

0.21

Interpreting Model Reasoning through Behavioral Counterfactuals

cs.LG cs.AI ·2026-10-22 ·Shawn Im, James Oldfield, Adel Bibi, Sharon Li
Abstract

Faithfully understanding the internal process that occurs as models form their chain of thought is an important problem with the rise in model capabilities. However, this remains challenging due to the variance in sampled responses, making it difficult to isolate the effects of interventions and compare model reasoning systematically. We propose a method for localizing and analyzing how a given intervention affects the model's reasoning through behavioral counterfactuals: given a fixed response, how does introducing an intervention to the prompt change the next-token distributions? We train a module to edit the residual stream such that it minimizes the KL divergence between the output distributions for a fixed response under a source prompt and its counterfactual target prompt. We find that matching the output distributions recovers the counterfactual behavior more effectively than standard activation reconstruction objectives and validate that attribution of the trained module outputs is effective for identifying causally important locations for counterfactual recovery. The attribution maps and visualizations of the downstream effects of model edits provide an interface for generating hypotheses about the original model's computation, which can then be tested with activation patching. In a case study on Gemma-4-E4B-it and DeepSeek-R1-0528-Qwen3-8B, we investigate how a suggestion influences model reasoning even when it is not verbally acknowledged. Our analysis identifies representations at the question statement and filler tokens that mediate a substantial part of the suggestion's influence, with downstream effects such as changing the framing of the justification or even the final answer.

0.21

The Geometry of Meaning in Neural Manifolds

cs.AI ·2026-10-22 ·Valentino Maiorca
Abstract · PhD thesis, Sapienza University of Rome. 214 pages, 62 figures, 32 tables. Advisor: Emanuele Rodolà. Co-advisor: Francesco Locatello (ISTA)

Just as different languages express the same concept through different words, neural networks express similar functions through incompatible coordinate systems. Different architectures, random initializations, and optimization trajectories produce incompatible parameterizations in extrinsic coordinates, obscuring functional relationships between independently trained models. When models independently solve similar tasks on related data, do their learned representations share exploitable structure despite incompatible coordinates? This thesis shows that semantic correspondence is a fundamental primitive revealing shared geometric structure beneath incompatible parameterizations. By identifying which elements refer to the same semantic content, correspondence breaks the symmetry of arbitrary implementation choices, anchoring incompatible systems to a shared semantic frame and revealing the intrinsic geometry that enables zero-shot transformation between independently trained models. Building on this principle, fourteen contributions explore correspondence-based alignment across zero-shot model stitching, cross-modal translation, policy transfer, activation steering, and neural decoding. They span domains (vision, language, reinforcement learning, neuroscience), architectures (CNNs, Transformers, graph networks), and scales (from small supervised models to billion-parameter generative systems). Each exploits different geometric properties (angular, metric, spectral, sparse, or distributional), yet all instantiate the same principle. It extends beyond artificial systems: fMRI data show that different human brains processing identical stimuli exhibit alignable geometric structure. By revealing intrinsic geometric structure beneath extrinsic parameterization, semantic correspondence transforms model isolation into interoperability.

0.21

Patronizing and Condescending Language Detection in English and Mandarin

cs.AI ·2026-10-22 ·Xinyi Vicky Xiang, Catherine Qingyun Wang, Paula Cortés Lemos
Abstract

Patronizing and condescending language (PCL) is a form of harmful language that belittles and assumes pity for vulnerable groups, reinforcing negative societal stereotypes. However, unlike overt hate speech, it is often difficult to detect, even for human judges. This paper proposes a system for the automated detection of PCL. Our first approach uses static word embeddings and a convolutional neural network for classification, while our second approach uses pre-trained BERT embeddings fine-tuned for the PCL task. Our second approach was considerably more successful, and achieved an F1 score of 0.528 over the development set.

0.21

CuraLoc: Repository-Level Fault Localization via Planned Code Inspection

cs.LG cs.SE ·2026-10-22 ·Sangeeta Warrier, Anush Sankaran, Ryen W. White
Abstract

A bug report describes what went wrong, but rarely where the code should change. Existing repository-level localization methods often let language models iteratively decide what code to inspect next, requiring repeated model calls and substantial token usage. We investigate whether these inspection decisions can instead be planned before source retrieval. In this research, we introduce CuraLoc, which first selects candidate files and represents their symbols using compact behavioral and structural descriptions. In a single planning decision, the model jointly selects starting symbols, chooses code relationships to inspect from each, and forms hypotheses about likely edits. Deterministic retrieval collects the requested source for a final ranking informed by those hypotheses, concentrating model reasoning on choosing and comparing code rather than directing each retrieval step. We demonstrate the efficacy of our proposed approach on the SWE-Bench Lite and Verified datasets. On SWE Bench Verified with GPT-5.2, CuraLoc raises Func@1 from LocAgent's 53.8% to 59.3% while using 90.3% fewer recorded tokens, and also outperforms RepoSearcher (56.2%) and Agentless (39.6%) on Func@1. On SWE-bench Lite, CuraLoc achieves a high 83.8% Func@5 accuracy improving on Agentless by more than 15% in relative accuracy. Additionally, CuraLoc supports different model backbones, with strong localization accuracy demonstrated using both GPT-5.2 and GPT-6 Astra. These results provide evidence that repository-level code inspection can be planned effectively without iterative model-controlled navigation.

0.21

DuoCook: An Overcooked-Inspired Benchmark for Visual Planning and Coordination

cs.AI ·2026-10-22 ·Ziyi Jiao, Heng Zhang, Drishti Regmi, Xiaokang Ye, Lianhui Qin +1 more
benchmark
Abstract

Coordinating two players in a cooking task requires a vision-language agent to track changing scene states, sequence recipe steps, and allocate shared resources under a time budget. We introduce an Overcooked-inspired benchmark for evaluating these capabilities in Unreal Engine environments constructed using coding agents. A single vision-language model controls both players from top-down visual observations, selecting high-level actions while the engine executes navigation. The benchmark contains 20 maps in five categories. These categories support evaluation of centralized coordination under equipment contention, changing terrain, and handoff dependencies. A prop legend and pot ingredient-count indicators provide explicit visual cues, while grid-based destinations support spatial grounding. The evaluation compares four VLMs on all 20 maps, measuring task score, order completion, delivery precision, action acceptance rate, and model API expenditure. A memory ablation on the same 20 maps examines the effect of a shared, self-authored notebook while retaining six-response action history in both conditions. The strongest evaluated model, \texttt{gpt-6-astra}, achieves 26.2\% mean order completion. Notebook memory has mixed score effects across models, with positive mean differences for Claude and Qwen and negative differences for GPT and Gemini.

0.21

xVGAE: A Hierarchical Variational Graph Autoencoder for Exchangeable Graphs

cs.LG stat.ML ·2026-10-22 ·Daniele Micheletti, Federica Zoe Ricci, Erik B. Sudderth
theory
Abstract

Graphs play a crucial role in many applications, ranging from drug discovery to social network modeling and astronomy. Generative models for graphs have substantially advanced in recent years, but we identify relatively simple scenarios where state-of-the-art models struggle to converge within a standard compute budget, or exhibit prohibitively slow generation times. We propose a generative architecture that achieves quality competitive with state-of-the-art models and is orders of magnitude faster at generation. Our approach is inspired by the Aldous-Hoover theorem, a classic representation theorem that characterizes any probability distribution of graphs that is invariant to permutations of node indices or exchangeable. This theorem establishes that three ingredients suffice: a graph-level latent variable, a set of exchangeable node-specific latent variables, and a so-called graphon function that maps graph- and node-level variables to the probability of an edge between any node pair. Given a training set of graphs, our exchangeable Variational Graph AutoEncoder (xVGAE) learns an approximation to their underlying graphons, as well as graph- and node-level latent variables. We show that our xVGAE encodes latent representations that correlate with graph- and node-level properties, and can (quickly, in one shot) generate new graphs that closely match key training set statistics.

0.20

Trace2World: Reconstructing App Worlds from Real Interaction Traces for Scalable GUI Agent Training

cs.LG cs.SE ·2026-10-22 ·Zezhou Wang, Hanzhang Zhou, Panrong Tong, Quyu Kong, Pengxiang Li +3 more
benchmark
Abstract · Preprint. Under review

Scaling GUI reinforcement learning requires environments that preserve real-app behavior. Yet coupling behavior inference with coding can turn partial traces into runnable mocks with incorrect state transitions. We introduce \textbf{Trace2World}, which separates global behavior reconstruction from implementation. It reconciles traces into a shared, source-linked app model, then compiles an executable environment. Independent replay of source workflows localizes mismatches for bounded implementation repair. Validated worlds then support learner-adaptive task and environment augmentation within source-supported behavior. From 1,570 traces, Trace2World constructs 50 environments. On 200 workflows across ten apps, 91.1% are reproduced end to end, versus 59.8% for the strongest matched task-local baseline. Source-grounded validation retains 94.0% of simulated training gains on real devices, versus 54.7% under self-validation. Learner guidance raises benchmark success from 56.9% to 61.1%. Training in Trace2World further improves the model by 20.2 points on our benchmark, 5.4--16.2 points across four external GUI benchmarks, and 16.3 points on 100 real-device tasks. Reliable reconstruction separates what behavior source evidence supports from whether the implementation preserves it.

0.20

Don't Forget What You Just Learned: Self-Directed Repair Turns Agent Mistakes Into Lasting Lessons

cs.AI cs.LG ·2026-10-22 ·Yuval Kansal, Karen Ullrich, Polina Kirichenko, Ilia Kulikov, Niraj K. Jha +2 more
Abstract

Intelligent agents should learn from their mistakes so that experience leads to continuous improvement. Yet today's agents recover from a mistake only temporarily: in-context reflections are easily forgotten across sessions and compete for limited context, while full-parameter fine-tuning is expensive and can interfere with existing capabilities. We introduce Self-Direction, a mechanism for agents to learn from mistakes that combines the benefits of temporary in-context retries and policy updates. When an agent fails a task, it reflects on its own trajectory in order to extract a lesson and immediately distills it into a lightweight low-rank adapter. By keeping the original weights frozen, Self-Direction prevents adapter training from overwriting base-model parameters while training additional parameters equivalent to ~0.02% of the base model's parameter count for each adapter. The agent then saves these adapters in a persistent library of lessons, automatically routing them to solve related problems later. Experiments with agents based on open-weight models from the Qwen family across open-ended research (GAIA), tool-use workflows (Claw-Eval), and terminal tasks (Terminal-Bench 2.1) show that our method improves overall success rates by 8.9 to 17.0 percentage points and performs competitively with frontier systems. We find that Self-Direction reduces repeated mistakes across future encounters of a task, nearly doubling the reliability of an agent's corrections. Learned lessons also transfer: a handful of adapters generalize to dozens of new tasks without additional training. Ultimately, this self-directed repair transforms everyday mistakes into lasting lessons.

0.20

PYTHIA, a transferable machine learning framework for simultaneous defect and adsorption site optimization

cs.LG ·2026-10-22 ·Ioannis Kouroudis, Zhaosu Gu, Poonam, Leon Muller, Alessio Gagliardi
Abstract

Machine learning has permeated material optimization, providing high value. Nevertheless, its re- liance on a high number of samples limits its applicability to relatively explored material families. In this paper we extend the applicability of machine learning to novel and understudied materials through the PYTHIA (Probabilistic Yet Transferable Hierarchical Interaction Architecture) framework. Pythia utilizes two of the most powerful machine learning concepts, foundational models and transfer learning. We employ a complex foundational model trained on a large dataset of chemically similar molecules, on a target that is physically close to our prediction target. Then, we freeze the weights and perform subspace alignment with the use of a transformer layer and add uncertainty quantification with gaussian processes. These two layers can be trained on a limited dataset yet allow us to leverage the gained knowl- edge and the studied materials and complex architectures. The resulting model is used as a surrogate for a Bayesian Optimization scheme. To showcase the robustness of the framework, we applied it to the investigation of adsorption and defect optimization of two distinct systems. Not only were we able to correctly and efficiently identify the optimal adsorption sites of even complex a

0.20

Same molecule, Different Knowledge: Cross-Identifier Binding Failures in LLMs For chemistry

cs.LG ·2026-10-22 ·Khouloud Saadi, Di Wang
benchmark
Abstract

Chemical AI workflows routinely represent the same molecule through multiple co-referring identifiers, including trivial names, IUPAC names, and SMILES, implicitly assuming that these representations provide interchangeable access to an LLM's knowledge. To test this assumption, we first introduce a controlled benchmark and show that it does not hold. Across foundation models spanning 7B to 72B parameters, even chemistry-specialized models, co-referring molecular identifiers consistently expose different subsets of the model's knowledge. We then conduct a series of controlled behavioral, representational, and causal analyses to characterize this failure. We show that even when a model can identify a molecule from its SMILES and retrieve a fact through the corresponding name, the same fact can remain inaccessible directly from SMILES. We term this phenomenon \emph{cross-identifier binding failure}. Further analysis of internal representations, conflicting identifiers, and parameter editing shows that neither improved chemical representation competence nor modification of individual facts is sufficient to establish reliable access across identifiers. Together, these results indicate that the core failure lies in the binding between an identifier and knowledge already present in the model. Motivated by this diagnosis, we introduce Binding Self-Distillation, a simple mitigation that uses lexical identity during training to transfer the model's existing knowledge access to the structural route. This enables factual retrieval directly from SMILES at inference time, without requiring a name lookup or training the underlying model from scratch.

0.20

From Prompt Rules to Golden Data Model: Production-Scale Generation of Personalized, Compliant Pharmaceutical Claims

cs.AI ·2026-10-22 ·Suyash Mishra, Qiang Li, Subhabrata Adak, Upender Phogat, Srikanth Patil +1 more
Abstract

Large language models (LLMs) are increasingly used to generate domain-specific commercial & medical content (e.g., claims, statements, and promotional phrases) in regulated industries such as pharmaceuticals. However, ensuring its regulatory compliance, factual accuracy, and audience-specific personalization remains a major challenge. We report a production case study asking whether moving compliance and personalization signal from prompt-level rules into retrieval time structured context improves pharmaceutical claims generation, using Composer, a deployed internal tool built on a Golden Data Model (GDM) of 1,867 pharmaceutical claims spanning four products. The GDM integrates claim metadata, references, audience labels, material type, usage history, Match Text Variant (MTV) links, and a six dimension quality score (6C). Across six product-audience segments, the context pipeline improves intended indication (+29%), intended content type (+62%) and intended target audience (+67%), 6C quality score (+86%), and claim reuse (+34%). Production log analysis over the deployment period shows a 29.5% reduction in weekly API cost. We report this as descriptive deployment evidence. This work is a systems and data architecture contribution: it shows that in a regulated, high-stakes domain, restructuring claim data for retrieval-time filtering is a more scalable lever than expanding prompt-level rules, and documents the engineering path data cleaning, graph linking, and quality scoring.

0.20

Track finding and particle identification performance with particle physics foundation models

cs.LG physics.data-anhep-exnucl-ex ·2026-10-22 ·Shuhang Li, Haiwang Yu, Gabor Galgoczi, Yeonju Go, Yi Huang +7 more
Abstract

Modern high energy nuclear and particle physics experiments produce large amounts of sparse data that must be reduced significantly in size before performing physics analysis. Recently, there has been an effort to evaluate Foundation Models as a potential tool for performing data analysis. These models have been shown to learn features in high dimensional datasets based on the multi-modal data they are trained on. To show their effectiveness at performing complex pattern recognition tasks, such as reconstructing charged particles in a magnetic field, we evaluated the performance of a Foundation Model based reconstruction on several downstream tasks with a dataset composed of spacepoints in a Time Projection Chamber. We show that the performance of the Foundation Model based data analysis significantly improves upon that of traditional methods, and furthermore can be adapted to learn additional tasks that are not necessarily related to the initial training objective.

0.20

Random Remasking Scales Masked Diffusion Inference

cs.LG ·2026-10-22 ·Zhaoyi Zhou, Frank Wu, Andrea Zanette
Abstract · 18 pages, 4 figures, 8 tables

Masked diffusion models (MDMs) generate sequences by progressively unmasking tokens, but errors introduced during decoding are typically irreversible. Recent work has shown that iterative correction can improve sample quality, yet existing theory fails to explain this given a learned, imperfect denoiser model. We re-formulate MDM inference as a Markov chain on the clean sequence space, defined by a remask-denoise kernel that randomly re-introduces masks and then denoises. This helps us analyze the actual denoising trajectory towards the data distribution. The analysis motivates Random Remask-Denoise (R$^2$D) inference, a purely test-time, finetuning-free, reward-free procedure. To further improve performance, we propose Trajectory Correction Training (TCT), a denoiser-training procedure that exposes the model to its own errors along the generation trajectory. On logical tasks, our approach exhibits a striking near-log-linear test-time scaling, achieving over 97% accuracy on hard Sudoku puzzles and over 99% accuracy on 3-SAT. Our approach also applies to pretrained models on actual language tasks, beating iterative correction baselines and the official Q-mode accuracy of LLaDA2.1-mini on HumanEval+.

0.20

Permission to Retry: A Study of Authorization and Recovery in AI Agent Workflows

cs.AI cs.SE ·2026-10-22 ·Deekshith Marla, Kushagra Bhatnagar
Abstract · 7 pages, 2 figures, 3 tables

A durable agent can resume a consequential action while its authority changes and the action's external outcome remains uncertain. We study execution-time reliability of tool-using AI agents, isolating recovery decisions from language-model behavior. Checkpoint state, permission to dispatch, evidence of an external effect and permission to reconcile are distinct obligations. We present a finite conformance study that checks these obligations through histories scored against a provider effect ledger hidden from the recovering application. A prospectively frozen campaign executes 336 histories on native LangGraph and Temporal hosts, crossing 14 schedules, three provider contracts and four matched application policies. All 168 paired host cells have identical checked outcomes. Per host, retained-ledger reconciliation passes all 42 histories: 21 checked completions, 15 denials and six unknown outcomes with reservations retained. Cached retry, revalidation before retry and stable-key retry pass 30, 36 and 31 histories respectively. A separate 16-case component study of InfinityOS, an enterprise AI platform developed by Arya.ai, exercises exported authority, worker, recovery and settlement code over repository fakes. The results isolate the roles of current authority, provider contracts and retained observations in recovery, with safety, completion and unresolved obligations reported separately for the declared application profile.

0.20

Building Presentation Intelligence for High-Quality Slide Generation

cs.LG ·2026-10-22 ·Yuhang Xie, Cheng Li, Chenliang Li, Mingyu Xu, Mingwei Sun +4 more
Abstract · 28 pages

PowerPoint presentations (PPTs) are a dominant medium for communication across diverse domains, yet high-quality PPT generation still depends heavily on labor-intensive human effort. Recent advances in large language models (LLMs) for agentic tasks have created both new opportunities and challenges (Chen et al., 2026) for automatic PPT generation. In this work, we present a unified training framework for automatic PPT generation toward presentation intelligence, integrating cross-validated data synthesis, guided teacher enhancement, and verifiable reward optimization. We first construct a cross-validated pipeline for data synthesis, with explicit control over both difficulty and validity. We then develop a teacher-enhancement framework grounded in rules and prior knowledge to substantially strengthen the instructional capacity of teacher models. Building on this, we further improve supervised fine-tuning (SFT) through a mixed-teacher strategy and a lightweight outcome-driven trajectory filtering method. Finally, based on the resulting SFT models, we introduce a suite of verifiable aesthetic metrics together with a GRPO-based reinforcement learning framework to optimize slide-level visual attributes beyond text-only supervision. Experiments show that our enhanced SFT pipeline improves Qwen-3.6-27B and Qwen-3.6-35B-A3B by 7.45 and 9.35 points on PresentBench, respectively, surpassing GLM-5.2 while maintaining strong generalization. Reinforcement learning further improves performance across five aesthetic dimensions, compensating for the limitations of SFT in visual refinement. Overall, the results demonstrate the effectiveness of our framework for high-quality PPT generation.

0.20

PhysTrace: Interpretable Temporal Evaluation of Physical Consistency in Video Generative Models

cs.AI cs.RO ·2026-10-22 ·Taekoan Yoo, Wonkyung Jung, Kyunghun Kim, Kyeongbo Kong
Abstract · 19 pages, 8 figures, 7 tables, including supplementary material

World action models increasingly rely on pretrained diffusion/flow-based video generators assumed to capture physical dynamics for prediction, planning, and action generation. Existing evaluations assess plausibility, probe representations, or measure predictive surprise, yet physical consistency is history-dependent: preceding events constrain what can physically occur later. We introduce PhysTrace, a training-free framework that uses MLLMs to formulate history-conditioned physical expectations and tracks their agreement with observed video windows through temporal video--language alignment, rather than relying on direct plausibility judgments or task-specific probes. On IntPhys2, PhysTrace achieves 67.2% possible--impossible discrimination accuracy, outperforming the evaluated direct MLLM baselines. We further introduce reference-guided masked denoising for controlled reconstruction from mixed possible--impossible video latents. To support these reconstructions and evaluate temporal localization as an additional capability, we introduce IntPhys2-Timing, adding violation intervals to IntPhys2, with PhysTrace achieving a Hit Rate of up to 77.4%. On these reconstructions, human evaluations reveal model-dependent differences in physical consistency and fewer physically impossible outcomes at higher initial noise levels, trends broadly captured by PhysTrace.

0.19

Transformers with CoT Discover Heavy-Ball Acceleration Through Training

cs.LG stat.ML ·2026-10-22 ·Yuxuan Cai, Yuan Cao
Abstract · 110 pages, 3 figures

Modern transformers often solve complex tasks by generating and refining intermediate reasoning steps. To develop a theoretical understanding of chain-of-thought (CoT) reasoning, a notable recent study by Huang et al. (2025) viewed transformers as algorithm executors, and showed that when transformers' CoT steps are supervised by a gradient-descent teacher, they can learn to implement gradient descent for in-context weight prediction in linear regression. In this paper, we study the same in-context weight prediction task but ask whether CoT reasoning can go beyond merely imitating the teacher algorithm. We show that, after initial supervision from a gradient-descent teacher, further trajectory-level supervision using the true regression weights leads the model to implement the Heavy-Ball method, thereby achieving acceleration over the original teacher. Notably, this accelerated algorithm emerges without supervision from any Heavy-Ball teacher. Our results provide a concrete theoretical example in which transformer reasoning can discover new algorithms through training, rather than merely imitate the algorithm used for supervision.

0.19

Bridging the Data Gap: Source-Informed Task-Space Transport for Few-Shot Diffusion

cs.LG cs.AI ·2026-10-22 ·Jinjian Jiang, Jing Peng, Rehan Deen, Yu Cao, Ziquan Liu +2 more
Abstract

When an object category is well represented in one domain but scarcely observed in another, adapting a diffusion model to the target can be unreliable. Yet the same source-target relationship may have been observed for other categories. We ask whether this recurring relation can be learned and reused for a new semantic task. We introduce Meta-Diff, which embeds source and target distributions in a shared low-dimensional task space, learns an affine transport map across paired training tasks, and generates from the transported source coordinate using a shared diffusion backbone. At deployment, it requires source examples and the identity of a previously learned relation, but neither target images nor target-specific optimisation. Gaussian-mixture experiments examine how transfer depends on task information retained by the encoder, relation predictability and the size of the source--target shift. On held-out CIFAR-100 validation tasks, correct-source transport achieves 13.9% lower mean sliced Wasserstein distance than wrong-source transport. In the one-target final-test comparison, it achieves 14.7% lower mean distance than target-only adaptation and 32.7% lower than full fine-tuning under validation-selected settings. These gains have clear limits: directly applying the known corruption to clean-source generations performs better on CIFAR, and requested-class identity remains weak. Office-Home shows domain alignment without reliable object identity; clinical stress tests show no resolved source-specific advantage. The results characterise when reusable task-space relations help and where their generative limitations remain.

0.19

Wavenumber-Aware Deep Learning for Robust Analysis of Raman Spectra

cs.LG physics.chem-phphysics.optics ·2026-10-22 ·Jihwan Kim, Jaeseong Heo, Yujin Kim, Sang Bin Yoon, Chanjoong Kim +5 more
benchmark
Abstract · 29 pages, 5 figures, 6 tables; Supplementary Information (15 pages) included

Machine-learning models for spectroscopy usually treat a spectrum as a fixed-grid intensity vector, leaving the physical coordinate implicit in the array index. Such models are tied to a single sampling grid, spectral range and resolution, which limits their use when measurement conditions change. Here we represent each spectral point as a pair of intensity and wavenumber and compute Transformer positional encodings from the wavenumber itself. On Raman spectra of three bacterial species and their synthetic mixtures, this representation remained robust to noise, sparser sampling and partial spectral coverage across convolutional, residual and Transformer networks. In controlled comparisons, wavenumber-based encoding outperformed index-based and learned encodings under partial observation and retained its performance on sampling grids never seen during training, where index-based encoding degraded. These findings were reproduced on a public 30-isolate bacterial benchmark. Explicit physical coordinates offer a simple route to robust spectral analysis under variable measurement conditions.

0.19

Reducing Multi-Horizon Photovoltaic Forecast Error through Tree-Anchored Expert Combination

cs.AI ·2026-10-22 ·Sun Ze, Zhou Liguo, Xu Yuqing, Yu Lei, Jiang Mingming +1 more
Abstract

A strong direct tree model may remain the best standalone photovoltaic forecaster, yet a weighted combination can reduce its residual errors. We examine whether complementary temporal predictors can lower the tree's mean absolute error (MAE) over 15--240 minutes without weaker ensemble members degrading performance. The candidate bank contains LightGBM and seven neural implementations with different temporal representations. A combiner fitted to MAE has nonnegative weights that sum to one. Unhelpful candidates can receive zero weight, while the tree anchors the combination and provides a development-stage fallback. The main design shares weights across horizons; horizon-specific weights and alternative losses test whether further complexity is needed. On 56,078 windows from two co-located public PVDAQ systems, the historical shared-MAE policy reduces scaled MAE by 3.21\% relative to the tree. The latest bounded neural correction does not reduce MAE further than this static combination. The results support selective combination of complementary predictions in this bank rather than equal averaging or indiscriminate addition of modules. Both analyses reuse a previously inspected evaluation period. They do not establish independent replication, weather-specific causation or universal gains from eight experts.

0.19

A Topology-Guided GeoAI Agentic Framework for Urban Flood Screening

cs.AI ·2026-10-22 ·Yutian Tao, Haoan Feng, Xin Xu, Leila De Floriani
Abstract

Open-world geographic artificial intelligence (GeoAI) agents can transform public geospatial data into accessible decision support for real-world challenges, but must balance autonomous reasoning with reliable spatial analysis and human oversight. We present a topology-guided GeoAI agentic framework for urban flood screening that takes a place name, drawn region, or uploaded boundary and returns ranked candidate basins with an interactive map and report. A deterministic controller retrieves public elevation, rainfall, soil, land-cover, exposure, and infrastructure layers, extracts terrain depressions with discrete Morse theory, and ranks basins with a relative flood-priority score. The language model acts only at recorded decision points: selecting a persistence threshold from a precomputed candidate table, drafting recommendations from computed evidence, and reviewing each recommendation against the same evidence. Three human-in-the-loop checkpoints let users revise the threshold, factor weights, or recommendation. In a College Park, Maryland case study, the selected threshold reduced 278 raw terrain minima to 70 basins, with 77% of these basins' lowest points also identified as sinks by a separately computed eight-direction (D8) flow-routing model. Across four further study areas, agreement ranged from 61% to 83%. By combining open data, explicit spatial reasoning, traceable evidence, and human oversight, the framework enables transparent first-pass flood screening while keeping consequential recommendations inspectable and user-controlled.

0.19

Group Relative Policy Optimisation Improves Multilingual Speech Recognition in Low-Resource Languages

cs.AI cs.CL ·2026-10-22 ·Tim Wenjie Hu, Evelyn Nafula Ouma, Benjamin Akera, John A. Quinn
Abstract · 5 pages, 3 figures, submitted to ICASSP 2027

Autoregressive encoder-decoder speech recognition models trained with teacher forcing can suffer from exposure bias, limiting their ability to recover from their own prediction errors at inference time. Recent work has shown that reinforcement learning with Group Relative Policy Optimisation (GRPO) improves speech recognition in major languages with LLM-based models. We extend this line of work to smaller, non-LLM encoder-decoder ASR models, using Whisper-large-v3 as a case study across 51 African languages. The benefits of this approach are most pronounced when training data is scarce or when input audio is noisy or degraded. Our resulting model, SunflowerASR, achieves the best published macro-averaged WER on the AfriVox-v2 (0.227) and SimbaBench (0.372) benchmarks, and is publicly available.

0.19

Visualizing and Interpreting Temporal Care Team Dynamics for Cancer Survival Risk Analysis

cs.LG ·2026-10-22 ·Yuhua Huang, Hsiao-Ying Lu, Kwan-Liu Ma
Abstract

Cancer care is a longitudinal, time-sensitive process requiring coordination among diverse healthcare professionals (HCPs), whose collaboration evolves throughout treatment. Prior studies link care team structure to patient outcomes, but the temporal dynamics of collaboration and their relationship to evolving patient risk remain poorly understood. We introduce a visual analytics framework for modeling and interpreting time-varying care team interactions in oncology. We construct dynamic collaboration networks from electronic health record (EHR)-mediated communication data and use a temporal graph neural network (TGNN) to estimate patient outcome risk over time. We then investigate how changes in care team composition and specialist participation relate to shifts in estimated risk across treatment stages. Our multi-view interactive system combines TGNN-based risk modeling with coordinated visualizations of care team networks, feature-level explanations from a surrogate logistic regression model, and exploratory what-if analysis. Without making causal claims, the system helps analysts examine how risk estimates co-evolve with collaboration patterns and identify care team characteristics associated with these changes. We demonstrate its utility through expert-guided analysis scenarios involving 505 patients with breast, colorectal, and lung cancers, evaluating its ability to reveal clinically meaningful collaboration patterns. This work improves the interpretability of temporal graph-based models in longitudinal, team-based healthcare, with potential applicability to other evolving collaborative settings.

0.19

Beyond Phenotypic Similarity: Relation Refinement and Selective Population Graph Learning for Multimodal Brain Disorder Diagnosis

cs.AI ·2026-10-22 ·Yinyu Lan
Abstract

Population graph learning supports brain disorder diagnosis by sharing information across subjects. Its effectiveness, however, depends on whether the relationships guiding this exchange are relevant to diagnosis. Similarity in demographic or acquisition attributes does not necessarily indicate diagnostic agreement, and heterogeneous subjects may require different representation updates within the same population graph. We propose PopDRP, a framework that combines diagnosis-guided relation refinement with hierarchical selective state-space learning. Using only training labels, Contrastive Affinity Refinement calibrates attribute-based affinities according to diagnostic agreement. Neural Graph Rectification then refines these relationships using fused functional-connectivity and non-imaging representations, guided by graph reconstruction and dual-view alignment objectives. On the resulting graph, a Selective Pyramid State-Space backbone integrates input-dependent state updates, graph diffusion, and hierarchical pooling and unpooling to learn population representations at multiple resolutions. Together, these components connect the refinement of inter-subject relationships with selective representation learning over the population graph. Experiments on ABIDE and ADHD-200 under stratified ten-fold cross-validation show improved accuracy and AUC over the compared baselines. Ablation studies further support the contributions of relation refinement and the selective hierarchical backbone to classification performance.

0.19

An Introduction to World Models for Generalist Agents

cs.AI ·2026-10-22 ·Zhihui Zhu
survey
Abstract

World models have recently emerged as a promising direction toward developing generalist agents capable of understanding, predicting, and interacting with diverse environments. However, despite growing interest, the concept of a world model remains loosely defined, encompassing approaches ranging from predictive dynamics models and learned simulators to models that jointly generate future observations and actions. This introductory paper aims to clarify the concept of world models and provide a unified perspective on their role in developing generalist agents. Starting from the partially observable Markov decision process (POMDP) framework, we connect world modeling to classical control, dynamical systems, signal processing, and reinforcement learning. We introduce different approaches to world modeling, emphasizing the knowledge they capture, their learning objectives, and how they support decision-making. We further explore the relationship between world modeling and policy learning, highlighting how these capabilities can be unified through conditional prediction without requiring separate models. Rather than offering an exhaustive survey of existing methods, this paper focuses on fundamental concepts, modeling principles, and connections, with the goal of providing an accessible and unified introduction to world models for generalist agents.

0.19

Muffin - A Shared Memory Buffer for Asynchronous AI Agents

cs.AI cs.MA ·2026-10-22 ·Ping-Lin Kuo, Suman Navaratnarajah, Christian Frey, Gresa Shala, Josif Grabocka
benchmark
Abstract

The rise of Agentic AI systems has changed the landscape of artificial intelligence by enabling strong reasoning capabilities through specialized multi-agent interaction with structured self-reflection. However, in target-driven reasoning tasks, single- and multi-agent systems are often on par in terms of performance gains. A major challenge is that failures often arise from an ineffective exchange of critical information during inter-agent interaction and coordination. Different from existing multi-agent debate paradigms that rely on synchronous turn-taking and sharing information across different agents, we propose an inter-agent protocol based on shared Memory bUFFer INference - referred to as MUFFIN - in an asynchronous agentic framework. In the first step, an orchestrator routes queries based on their reasoning demands and delegates demanding ones to a multi-agent setting, where a shared memory buffer provides asynchronous exchange of intermediate findings while an agent's private reasoning remains local. Our experiments show that our framework Muffin outperforms established competitors, achieving the best accuracy in 17 out of 20 of our benchmark settings, and provides a better quality-cost trade-off than multi-agent debate: Muffin improves accuracy by 19.2 accuracy points while using 77.4% fewer tokens, without requiring full-trace sharing.

0.19

Multi-Token Prediction and Alternative Tokenization Improve Learning Latent Multi-Hop Graph Search at Scale

cs.LG cs.CL ·2026-10-22 ·Arvid Frydenlund, FuTe Wong, Katherine Lambert, Frank Rudzicz
Abstract · Don't be intimidated by the page length :)

Given an unseen graph, a start node, $s$, and an end node, $t$, can a language model generate the shortest path from $s$ to $t$? Prior work has found that models fail to learn this task as the graphs scale. We show that this failure has two causes: the supervision signal and the tokenization. First, we theoretically establish and empirically demonstrate the effectiveness of multi-token prediction (MTP), and show that the benefits of MTP are topologically agnostic. Second, we introduce alternative tokenizations and show that the two causes are orthogonal. Using these methods, we break through the scaling wall observed by prior work, consistently and stably learning to solve the task at scales far beyond previous limits.

0.18

Adaptive Skill Granularity for Coding Agents: Runtime Routing and Macro-to-Micro Recovery

cs.AI cs.SE ·2026-10-22 ·Dingxuan Zhang, Yanhui Jiang, Jun Wang, Hefeng Zhou
Abstract

Reusable skills allow large language model agents to package task-specific procedures, tool affordances, and execution guidance into modular units. Prior work has largely studied how agents retrieve, compose, or synthesize skills before execution. In coding agents, however, execution poses an additional runtime control problem: the same task may be solved through primitive tool calls, macro skills, or hybrid execution that starts with a macro and falls back to micro-level recovery. We introduce Adaptive Skill Granularity, a framework that treats skill granularity as an action-abstraction policy conditioned on task structure, execution state, and macro-skill failures. The system combines a feature-based granularity controller and a finite-state recovery executor. We evaluate the framework with a deterministic coding-agent protocol using verifier-defined success, total cost, and process metrics such as macro adoption, failure recovery, route match, and transition correctness. Results show that fixed macro execution is efficient when applicable but brittle under mismatch, while runtime routing and explicit macro-to-micro recovery improve robustness and recover controlled macro failures that fixed-granularity execution leaves unresolved. Our findings suggest that skill libraries for coding agents should be treated not only as retrievable resources, but also as controllable, verifiable, and recoverable execution abstractions.

0.18

Reliable AI-Assisted Decisions via Selective Auditing

cs.LG stat.MLstat.ME ·2026-10-22 ·Isaac Gibbs, Elise L. Han, Ryan J. Tibshirani, Tijana Zrnic
Abstract

We consider online monitoring of an AI system that produces a sequence of decisions. At each round, the system outputs a proposed decision together with a confidence score estimating the trustworthiness of the decision. The decision and confidence score need not be produced by the same model: for example, one agent may propose a decision, while a separate verifier agent evaluates it and outputs a correctness or safety score. Based on this information, we must decide whether to accept the AI decision automatically or consult an expert, such as a human auditor, who reviews the proposal and either accepts or corrects it. We formulate this task as an online constrained optimization problem in which the goal is to minimize the frequency of expert auditing while guaranteeing that the fraction of AI errors that go unaudited stays below a user-defined threshold. We derive a family of algorithms with provable error-control guarantees with no assumptions, including a method that is optimally efficient: it asymptotically attains the best achievable auditing rate for both adversarial and i.i.d. data streams. Experiments on screening harmful requests to large language models, qualitative text annotation, and data labeling show that our methods achieve valid error control in practice while substantially reducing the need for human auditing compared to baselines.

0.18

Training Mixed-Monotone Neural Networks for Verification Without Set Propagation

cs.LG ·2026-10-22 ·Jonas Fingerhut, Lukas Koller, Matthias Althoff
theory
Abstract · 20 pages, 3 figures, 9 tables

Most formal verification approaches for neural networks conservatively enclose the output set of the neural network. However, computing a tight output enclosure is in general computationally hard. We sidestep this challenge with a novel neural network architecture that realizes mixed-monotone functions as the difference of monotone subnetworks. By utilizing the theory on mixed-monotone systems, this reduces the computation of a tight output enclosure from complicated set propagation to two simple forward passes of opposite corners of an input interval. Moreover, because the size of the resulting enclosure is a closed-form function of the weights of the neural network, we can directly minimize it during training. We prove a lower bound on the size of any such enclosure, given by the total variation of the represented function generalized to several inputs, which makes it possible to compare the tightness of the enclosure against the optimum. Our evaluation on regression and dynamical-system benchmarks demonstrates that our training reduces the enclosure size to within 0.5% of the lower bound without sacrificing prediction accuracy. Computing our enclosure is orders of magnitude faster than CROWN and $α$-CROWN and on deep networks it is also tighter than zonotope propagation and CROWN, while its size exceeds that of $α$-CROWN by only 5.1%.

0.18

ENCL: Enabling Efficient Neuromorphic Continual Learning with Local Learning Rule and Sub-Network Freezing Strategy

cs.NE cs.AIcs.LG ·2026-10-22 ·Rachmad Vidya Wicaksana Putra, Shuochen Zhou, Hassan Raza, Muhammad Shafique
Abstract · 9 pages, 6 figures

Continual learning (CL) in neural networks is a promising learning approach for updating the knowledge of deployed systems based on new data encountered at run time, that have not been learned during the offline training phase. To enable their power/energy-efficient CL processing, neuromorphic continual learning (NCL) concept with sparse event-based operations have recently been introduced. However, state-of-the-art NCL methods typically employ the memory-hungry data/experience replay, coupled with costly global learning rule (i.e., backpropagation-based surrogate gradient). To address these limitations, we propose ENCL, a novel methodology that enables efficient NCL by employing a lightweight local learning rule and sub-network freezing strategy. ENCL leverages the temporally- and spatially-local learning rule to perform efficient credit assignment for training new tasks, alleviating the expensive cost of global learning rule. ENCL also employs a sub-network freezing strategy based on the given tasks to avoid the needs for network architecture expansion, hence further improving the efficiency and scalability of the NCL training process. To efficiently support this, we train each task with a small number of epochs as well as adjusted learning rate and timestep. Experiment results on task-incremental learning (TIL) scenario show that, our ENCL consistently achieves strong performance across different datasets as compared to state-of-the-art works. Specifically, ENCL achieves competitive accuracy comparable to that of the state-of-the-art data/experience replay, while reducing memory cost by up to 88%, NCL training time by up to 51%, average power consumption by up to 40%, and energy consumption by up to 65% across different benchmarking datasets (i.e., Split MNIST, Split Fashion MNIST, Split CIFAR-10, and Split CIFAR-100).

0.18

Jev-DLC: Decision-Level Continual Learning from Typed Feedback

cs.AI ·2026-10-22 ·Yangyang Feng, Jianle Gan, Zhuoyan Feng, Chengwei Qin
Abstract

Individual preferences often become visible only after deployment, when a user accepts one decision or revokes it in favor of another. Jev-style typed interfaces make this feedback explicit by identifying the selected action and, under revocation, the decision it replaces. We study how these consequences persist across requests as Decision-Level Continual Learning from Typed Feedback (Jev-DLC). Jev-DLC treats each correction as both evidence about a recurring action identity and a directed relation between the preferred and rejected choices. It preserves these signals in complementary bounded channels above a frozen scorer and combines their scores to shape the next decision. The method retains no raw interaction history, adds no scorer call, and performs no runtime language-model backward pass. We also introduce TCPA-Bench, a chronological suite that credits a correction only through its effect on later decisions. On TCPA-Repeat, Jev-DLC reaches 44.83% user-macro accuracy overall, 2.43 points above Recent-4 history replay. After at least two earlier feedback events, it reaches 56.07%, a 7.54-point margin over the strongest reported comparison at this depth. These results show how explicit typed corrections can become compact causal state for continual personalization.

0.18

Predictable performance is suboptimal, and what it means for model routing

cs.LG cs.AI ·2026-10-22 ·Guanhua Zhang, Moritz Hardt
Abstract

Large language models (LLMs) differ in capabilities and costs, motivating routing methods that adaptively choose a model for each query. These methods rely on performance prediction: given a query, estimate whether a model will answer it correctly. Most routing methods train separate predictors of candidate models' performance to guide model selection. Our work shows that performance prediction only beats a simple baseline when the candidate model is suboptimal on the task or the performance predictor adds significant capacity relative to the candidate model. Specifically, we take a model's self-confidence, the probability it assigns to its chosen answer, as a baseline and study how much a performance predictor can improve on it. First, we show that any performance predictor that outperforms a model's self-confidence can be used to improve the model's loss: the predictor's advantage transfers exactly in log loss through recalibration. Second, accounting for the predictor's additional parameters, we bound its advantage by the model's suboptimality plus the gain from those parameters. Third, we extend our theoretical results to two-model cascades, which route queries to a more capable model when the first model is predicted to fail. Experiments with 16 language models on four benchmarks show that predictors gain a larger advantage over self-confidence on more suboptimal models. This advantage shrinks as models improve through finetuning. Cascading exhibits the same two patterns: predictor-based routing offers larger gains over confidence-based routing for more suboptimal models, and these gains diminish after finetuning.

0.18

Adaptive Deadline- and Energy-Aware Caching Algorithms in Dynamic Multi-UAV Edge Networks

cs.LG cs.NI ·2026-10-22 ·Jian Kim, Ki-Il Kim
Abstract · 18 pages, 7 figures

Content caching via unmanned aerial vehicles (UAVs) becomes one of the solutions to maintain content delivery when terrestrial infrastructure is unavailable or disrupted. However, existing caching approaches often focus primarily on cache-hit performance or average delivery delay, without jointly addressing spatially heterogeneous and time-varying demand, content-specific deadlines, and uneven communication-energy consumption across UAVs. To address these challenges, we formulate a dynamic deadline- and energy-aware cache placement and replacement problem that jointly considers deadline misses, UAV-side communication energy, and inter-UAV energy imbalance. To achieve this, we first develop deadline- and energy-aware UBAC (DE-UBAC), an interpretable hierarchical rule-based realization that incorporates deadline-aware prioritization, cooperative placement, energy-aware coordination, and cache-update cost. However, predefined priorities and thresholds limit its adaptability under dynamically coupled network conditions. We therefore propose GRU-assisted Deadline- and Energy-aware Proximal Policy Optimization (GRU-DEPPO), which combines local and global GRU-based temporal representations with a centralized PPO actor-critic for adaptive cache placement and replacement. Simulation results show that GRU-DEPPO achieves the lowest deadline-miss rate and total UAV-side communication energy among the considered schemes. Compared with DE-UBAC, GRU-DEPPO reduces the deadline-miss rate by approximately 61% and inter-UAV energy imbalance by approximately 40%.

0.18

Shift-aware score transport for privacy auditing without retraining

cs.LG ·2026-10-22 ·Yaohong Yang, Gauri Pradhan, Qi Chen, Samuel Kaski, Antti Honkela
Abstract

Privacy auditing empirically assesses privacy leakage by using membership inference attacks to establish lower bounds on privacy loss. We consider an observational auditing setting in which the auditor cannot insert canaries, control data inclusion, or retrain the target model, and must instead rely on fixed sets of known members and non-members. This setting is relevant for large deployed models or when retraining is restricted by computational or access constraints. In this case, non-members available for auditing may differ in distribution from the training data, causing membership-inference scores to be skewed due to the distribution shift. We address this problem by estimating the counterfactual in-distribution non-member score distribution using the available out-of-distribution (OOD) non-members. Our proposed method learns the relationship between data representations and membership-inference scores from OOD non-members and transports it to the target covariate distribution. Because this transport may become unreliable under extrapolation or imperfect conditional invariance, we further introduce a conservative correction calibrated directly to the nominal transported privacy level. We construct controlled calibration tasks from the available non-member data to estimate how representation-space displacement inflates the audit, and use this estimate to correct the transport-induced audit error. Across multiple datasets with distribution shifts, our method yields substantially more stable privacy estimates while retaining tighter lower bounds than baselines.

0.17

RiboBridge: protein-conditioned de novo RNA binder structure design

cs.LG q-bio.BM ·2026-10-22 ·Zhou Zhang, Hanqun Cao, Minghao Sun, Odin Zhang, Yang Zhang +3 more
benchmark
Abstract

Designing RNA molecules that bind to target protein structures requires generating both an RNA conformation and its bound complex geometry, without assuming a predefined binding site. Existing approaches generally model these coupled processes within a unified representation, entangling RNA conformation generation with protein-relative placement. In this work, we introduce RiboBridge, a structure-first framework that separates RNA conformational generation from protein--RNA assembly through cross-frame interaction geometry. RiboBridge first proposes multiple candidate RNA-binding interfaces on the protein surface, generates canonical RNA conformations within a pretrained RNA structural latent space (RiboSphere), and then predicts protein--RNA interaction constraints to bridge the independently defined RNA and protein coordinate frames. The resulting interaction geometry is converted into a geometrically consistent protein-relative pose through rigid placement. We evaluate RiboBridge on a temporally split, sequence-filtered, and redundancy-controlled benchmark of 124 protein--RNA complexes. Under post-hoc interface-correspondence evaluation, where native information is used only after generation to select among already generated predicted-pocket candidates, RiboBridge achieves the highest fNAT and DockQ and the lowest interface and ligand RMSDs among compared methods. Component ablations and stage-wise diagnostics further characterize the contributions of RNA conformational generation, interaction prediction, and rigid placement, identifying the main sources of complex-recovery error.

0.17

Rewriting Scaling Laws for Synthetic Data with Compute and Data Constraints

cs.LG cs.CL ·2026-10-22 ·Nicholas Roberts, Ahmed Ahmed, Suhas Kotha, Percy Liang, Yu Wang +2 more
Abstract

Synthetic rewrites, where a language model rewrites a natural document e.g. into a target style, are now a standard ingredient of pretraining corpora. However, critical aspects of their use remain poorly understood, such as how much compute should be spent rewriting training data and whether a model trained on rewrites can outperform the rewriter. We study these two questions by proposing a rewriting scaling law that predicts loss from model size, training tokens, the fraction of training tokens that are rewrites, and the number of epochs over the natural data, charging the compute spent generating rewrites. We fit this scaling law in two settings. In the compute-limited setting, natural data is plentiful and rewrites come from a third-party rewriter trained with more compute than our largest run. We find that the estimated optimal rewrite fraction increases with compute, from under $10\%$ at $3\times10^{18}$ FLOPs to $32\%$ at $10^{22}$ FLOPs. Our scaling forecast extrapolates $33\times$ in FLOPs to train a 22B-parameter model, whose final loss is within $1.1\%$ of our prediction. This model achieves a better BPB than its rewriter on several web domains at one-tenth the rewriter's training compute. In the data-limited setting, we assume we only have a fixed seed set of natural data and that our rewriter is a model of the same size as the target, pretrained on that seed set. Again, we find the optimal rewrite fraction increases with the compute budget. We also demonstrate that a model trained on rewrites of its own pretraining data, generated by a rewriter of the same size, beats the best run on natural data alone at every model size up to 3B. Our results show that synthetic rewriting can be studied through the scaling law lens: that optimal methodological choices are predictable and scaling does not necessarily break down beyond the rewriter's capabilities.

0.17

Large language models show systematic failures in incorporating evolving medical knowledge

cs.AI ·2026-10-22 ·Min Zeng, Zaifu Zhan, Kai Yu, Rui Zhang
Abstract

Medical knowledge evolves, yet whether large language models (LLMs) faithfully incorporate revisions into clinical recommendations remains unclear. We introduce ChronoEvo, comprising 1,709 recommendation-level revisions across 12 forms of medical knowledge change and 275 guideline lineages, to distinguish recognition of revised guidance from its full incorporation. With updated guideline evidence explicitly provided, revision recognition exceeded full incorporation across 14 models, with a 20.1-percentage-point gap in the strongest model. For Qwen3-4B, 90.8% of responses recognized revisions, 80.2% were clinically correct, and 58.5% fully incorporated them, yielding a 32.3-percentage-point gap. Two clinicians' adjudicated assessment of 300 responses yielded a 40.3-percentage-point gap in this enriched validation sample; 61 of 144 responses with incomplete incorporation were clinically incorrect. Prompting Qwen3-4B to prioritize revised evidence reduced outdated-content retention and improved clinical correctness without significantly improving full incorporation. These findings identify faithful incorporation of revised guidance as a distinct challenge for clinical LLM reliability.

0.17

GIFT: Gradient Informed Filtering for Distillation Augmented Reasoning RL

cs.LG ·2026-10-22 ·Shijie Li, Yilin Gao, Tieyuan Chen, Yun Luo, Weiyao Lin +1 more
code/data
Abstract

Training large language models for complex reasoning tasks relies on two complementary signals: sparse, outcome-grounded reinforcement learning (RL) for correctness, and dense, token-level distillation for guidance. However, complementarity in supervision does not guarantee compatibility in optimization; imitation-focused distillation can conflict with the outcome-driven RL objective, and existing hybrids lack a principled resolution. We propose \textbf{GIFT} (\textbf{G}radient-\textbf{I}nformed \textbf{F}il\textbf{T}ering for Distillation-Augmented Reasoning RL), a novel method that directly addresses the allocation problem of which teacher updates to apply. GIFT performs a token-level geometric analysis, decomposing the distillation gradient into a component \textbf{parallel} to the RL gradient, which measures first-order conflict and provides a \textbf{safety} filter, and a component \textbf{orthogonal} to it, which quantifies novel directional guidance and provides a measure of \textbf{value}. This decomposition yields a principled scoring rule to rank token-level guidance. We combine this with a progressive \textbf{coverage annealing} schedule that shrinks the budget of tokens receiving distillation, ensuring that the most conflicting updates are discarded first while the most complementary ones are retained at full strength. This facilitates a smooth transition from dense, geometrically-filtered guidance to a pure, outcome-grounded objective at convergence. Extensive experiments on mathematical and logical reasoning benchmarks show that GIFT consistently outperforms strong baselines across multiple model scales, achieving relative gains of up to 9.5\% on difficult mathematical reasoning tasks. Code: https://github.com/lilifree/GIFT

0.17

Phase Transition from Random Features to Fully Trainable Networks

cs.LG math.NAmath.STstat.ML ·2026-10-22 ·Xin He, Huan Lao, Xiliang Lu, Shijun Zhang
Abstract

Constructing accurate and computationally tractable approximation spaces is a fundamental task in scientific computing. Partially trainable neural networks bridge fixed random features and fully adaptive representations. This paper quantifies a phase transition pattern in approximation bounds and observes analogous behavior in training errors: a rapid but smooth initial improvement followed by diminishing returns, with a sharper transition under greater initialization mismatch. Under polynomial spectral decay with exponent $α>1$ and leverage compatibility, we establish a squared $L_2$ approximation rate $\mathcal{O}((κ_P\log(m/δ)/m)^α)$ for two-layer networks of width $m$ and trainable proportion $P$, where $κ_P=(P/κ_1^*+(1-P)/κ_0^*)^{-1}$. The bound holds uniformly over the unit ball of a reproducing kernel Hilbert space (RKHS) with probability at least $1-δ$. The constants $κ_0^*$ and $κ_1^*$ quantify the fixed and trainable groups' sampling mismatches relative to the target RKHS. For spherical biased-ReLU features in dimension $d$, we verify these assumptions with $α=(d+2)/(d-1)$ and $κ_0^*=4\sqrt{2}τ(d+1)/d$ for initialization scale $τ\ge1$, and $κ_1^*=1$ under ridge leverage sampling. For $d=5$ and $τ=10$, a trainable proportion of approximately $4.1\%$ captures $90\%$ of the reduction in the approximation bound from $P=0$ to $P=1$. A structured three-layer example illustrates how a few trainable first-layer neurons exploit fixed downstream blocks through nonlinear composition. Numerical experiments across diverse settings, including networks with up to 15 layers, exhibit similar phase transitions in training errors.

0.17

Nonparametric autoregressive estimation of probability distributions with parallel gradient boosting

cs.LG stat.ML ·2026-10-22 ·Rémy Chapelle
Abstract

Nonparametric estimation of joint probability distributions is a core problem in statistics, which is notoriously challenging with high-dimensional data. Such challenges also arise with conditional distribution estimation and have led to flexible algorithms that aim to tackle the ''curse of dimensionality'' by exploiting predictive structures of lower dimension than the ambient space. In this paper, we study the use of such conditional distribution estimators to estimate joint distributions by factorizing the latter in an ''autoregressive'' way. We show that for loss functions satisfying a subadditivity property, bounds on the convergence rate of an autoregressive estimator can be derived from bounds on the convergence rate of the conditional distribution estimators. These bounds relate the difficulty of estimating the joint distribution to that of estimating the most challenging factor in the autoregressive factorization. When the conditional estimators adapt to sparsity in their conditioning sets, we characterize this difficulty through variable elimination in an undirected dependence graph. This allows bounding the rates of convergence by quantities related to the sparsity of the graph, and provides an additional motivation for the autoregressive approach. We then evaluate the practical performance of an autoregressive estimator based on parallel gradient boosting, a flexible nonparametric method for estimating conditional distributions. The estimator demonstrates competitive estimation performance on various simulated and real datasets, even without optimizing the factorization order. Together, these results support nonparametric autoregressive estimation as a practical approach to flexible joint distribution estimation.

0.17

Cautious Conflict-Free Inverse Gradient Methods

cs.LG ·2026-10-22 ·Damian Hattler, Qiang Liu, Nils Thuerey
code/data
Abstract · Code available at [https://github.com/damiha/CautiousConFIG](https://github.com/damiha/CautiousConFIG)

Partial differential equations (PDEs) are indispensable tools for modeling complex phenomena in physics and engineering. Physics-Informed Neural Networks (PINNs) have emerged as a powerful alternative to traditional numerical solvers by directly incorporating physical constraints into neural network loss functions. Despite their advantages, PINN training requires multi-objective optimization, which often leads to significant gradient conflicts among boundary, initial, and residual loss terms. The Conflict-Free Inverse Gradient (ConFIG) method addresses these conflicts; however, integrating ConFIG with adaptive optimizers such as Adam introduces a critical vulnerability: the conflict-free direction generated by ConFIG is further processed by the optimizer, and the resulting weight update may no longer be conflict-free with respect to the task gradients. To address this issue, Cautious ConFIG is proposed as a concise three-line extension to the ConFIG algorithm that empirically restores the conflict-freeness guarantee prior to the weight update. The method is evaluated on three PINN benchmarks: the 1D Viscous Burgers' equation, the 1D Nonlinear Schrödinger equation, and the 2D Kovasznay flow. Experimental results demonstrate consistent performance improvements over the standard ConFIG baseline. Additionally, the stabilizing effect of cautious masking and its reaction to conflict severity, task number, and optimizer lag are investigated using a simple toy system.

0.17

SQuAT: Spherical Lattice Vector Quantization-Aware Training

cs.LG cs.CL ·2026-10-22 ·Sebastian Loeschcke, Ivan Koryakovskiy, Tycho F. A. van der Ouderaa, Riccardo Del Chiaro, Mart van Baalen +2 more
Abstract · 25 pages, 7 figures

Vector quantization (VQ), in particular lattice-based methods, offer strong rate-distortion trade-offs for low-bit large language models (LLMs), yet their behavior under quantization-aware training (VQ-QAT) remains poorly understood. We scale lattice VQ-QAT to billion-parameter LLMs and multi-billion-token training regimes, and introduce SQuAT, a shape-gain VQ-QAT approach trained with a separate straight-through estimator (STE) for the shape and gain. SQuAT decomposes weights into direction and magnitude: normalized directions are quantized using an $E_8$-derived spherical codebook, while gains are quantized independently with low-bit scalar quantizers. We show that, unlike a standard STE applied to the full quantizer, which ignores the shape-gain structure, these separate STEs induce tangent-space updates for directions, yielding an implicit form of spherical (Riemannian) optimization. To enable scalability, we develop a distributed exact nearest-neighbor assignment scheme, in which workers quantize disjoint weight blocks and communicate only discrete indices, reducing VQ-QAT training time by 12.8$\times$ and allowing us to scale it to 10B tokens. Across instruction-tuned LLaMA-3.2-1B/3B, Qwen3-4B, and base Qwen3-1.7B, SQuAT consistently outperforms standard VQ-QAT STE under identical quantizers and initialization, and achieves SOTA VQ-QAT accuracy.

0.17

Efficient Omni-modal Large Language Model Inference via Per-Query Modality Selection

cs.LG ·2026-10-22 ·Shuqi Liu, Yuzhou Cao, Luke Ong
Abstract

Omni-modal large language models (Omni-LLMs) can jointly process a broad range of modalities, such as text, audio, and video, enabling stronger multimodal understanding and reasoning capabilities, but often at substantial inference cost. We observe that the benefit of additional modalities varies across queries, indicating that a fixed modality configuration can be inefficient. Motivated by this, we study query-wise modality selection, which keeps the underlying Omni-LLM fixed while adaptively selecting which modalities to provide for each query. We formulate the problem as an accuracy-cost optimization problem and develop a cost-sensitive learning approach for modality selection. We evaluate our approach on WorldSense using Qwen2.5-Omni and Qwen3-Omni under three inference-cost measures: input tokens, latency, and monetary cost. Across both models, query-wise modality selection consistently improves the accuracy-cost trade-off and can even outperform the standard setting in which all available modalities are provided to the Omni-LLM. For Qwen2.5-Omni, our cost-sensitive approach matches or exceeds this full-modality accuracy while requiring only approximately 41% of the input tokens, 39% of the latency, and 61% of the monetary cost.

0.17

REDAR: Reinforcement-Driven Adaptive Regularisation for Class Incremental Learning

cs.LG ·2026-10-22 ·Akila Ayanthi, Darshana Priyasad, Tharindu Fernando, Sridha Sridharan, Clinton Fookes +1 more
position
Abstract · The 28th International Conference on Digital Image Computing: Techniques and Applications (DICTA 2026)

Regularisation-based class-incremental learning (CIL) methods control the stability-plasticity trade-off through fixed coefficients tuned offline, independently of the continual learning objective and blind to phase-specific variations in data volume, task difficulty, and class similarity. We argue this is a fundamental mismatch and propose REDAR: Reinforcement-Driven Adaptive Regularisation for Class Incremental Learning, which treats regularisation strength selection as a sequential decision problem. At each incremental phase, a lightweight reinforcement learning (RL) agent observes a compact state of current performance metrics and incoming data statistics, and outputs the distillation coefficients for that phase. To respect CIL's constraint on historical data access, the policy is trained entirely on pseudo-CIL tasks constructed from the initial phase, then deployed on the incremental sequence without further adaptation. Experiments on CIFAR-100 and Tiny-ImageNet demonstrate that REDAR reduces forgetting and improves forward transfer over fixed and heuristic baselines, and that policies trained on short pseudo-task horizons transfer effectively to longer incremental runs.

0.17

METTA: Memory-Enhanced Test-Time Adaptation for Meta-Reinforcement Learning

cs.LG cs.AI ·2026-10-22 ·Ali Najar, Hosein Hasani, Amir Kooshan Fattah, Yasamin Rajabi, Mahdieh Soleymani Baghshah
Abstract · 9 pages

Meta-reinforcement learning aims to train agents that adapt quickly to unseen tasks from limited interaction, but this requires representing task evidence over long interaction histories. Classical recurrent agents compress experience into hidden states, while transformer-based agents retrieve from past trajectories at quadratic memory and computation cost. In this work, we propose Memory-Enhanced Test-Time Adaptation (METTA) for meta-reinforcement learning. It decomposes the fast adaptation phase into two temporal scales: a causal neural memory performs fine-grained within-episode adaptation with linear sequence complexity, while compact episode-level summaries support long-range cross-episode task inference without attending over the full interaction history. Moreover, equipping agents with neural memory provides sufficient expressive capacity and test-time training capability, enabling effective adaptation to unseen held-out tasks and task families at meta-test time. On unseen held-out Meta-World ML10/45 test environments, METTA achieves a success rate over $50\%$ while maintaining a low training budget and short wall-clock time. On long-context sparse-reward PointMaze tasks, it outperforms transformer-based baselines by exhibiting more structured exploration behavior. These results show that METTA provides an efficient path toward more adaptive and scalable long-context meta-RL.

0.16

EEG-Captioner: Grounding EEG-to-Language Generation in Fundamental Signal Properties

cs.LG ·2026-10-22 ·Leyuan Jin, Jiahao Mei, Yixuan Yuan, Shimeng Li, Mengyue Wu +1 more
Abstract · 5 pages, 2 figures, 3 tables. Submitted to ICASSP 2027

Electroencephalography (EEG)-to-language models offer a promising approach for translating complex neural signals into human-readable descriptions. However, generating plausible language does not necessarily demonstrate that a model has learned to interpret the underlying EEG signal. A direct way to assess such interpretation is to probe objectively measurable properties of the signal itself. We therefore construct an EEG-language dataset of 160,781 segments annotated with signal-level question answering (QA) pairs, where each question targets one of five property families and requires a quantitative comparison involving spectral, spatial, temporal, or lateral signal properties. Each segment is also paired with a feature-grounded EEG caption. We further propose a Dual-Domain EEG Encoder that jointly models temporal waveforms and spectral-spatial representations, and introduce EEG-Captioner, a unified EEG-to-language framework trained with both signal-level QA and EEG caption supervision. Experiments show that the proposed encoder achieves the highest overall macro-averaged balanced accuracy (Macro-BA) among the compared EEG encoders, while EEG-Captioner achieves a Macro-BA of 88.81% on signal-level QA and improves caption factual consistency from 82.06% to 83.46% over caption-only language adaptation, demonstrating the complementary benefits of signal-level QA and caption supervision. Demo page: https://lyJinAoki.github.io/EEG-Captioner/.

0.16

Questioning Instrumental Convergence in RLVR: Checkpoint Evidence from OLMo-3

cs.LG cs.AI ·2026-10-22 ·Barış Alparslan, Sandy Tanwisuth, Caleb Petterson
benchmark
Abstract

Instrumental convergence is often invoked as a general reason to expect capable agents to acquire or express power-seeking subgoals: self-preservation, resource acquisition, deception, or resistance to oversight which poses an important threat to the society. For LLMs, reinforcement learning is the first stage of post-training that directly optimizes behavior against an objective; therefore, it is the stage where convergent instrumental behavior, on IC grounds, likely to appear. Does reinforcement learning with verifiable rewards (RLVR) make a model more likely to express instrumental-convergence behavior, or does this depend on the training regime, evaluation frame, and stage of post-training? We study this question as a checkpoint-conditioned behavioral trajectory rather than as a single before/after safety score. Using InstrumentalEval, we evaluate OLMo-3-7B-Think across SFT, DPO, 13 intermediate RLVR checkpoints, and the final RLVR model. The trajectory is nonmonotonic: instrumental rate rises from 25.0% at SFT to 46.1% near RLVR step 200, then falls to 11.8% at the final checkpoint. Paired item-level tests and step-wise regression support the final SFT-to-RLVR reduction. Prompt framing does not explain this pattern: a harmful-prompt indicator combined with other framing effects explains only 15% of within-checkpoint score variance on average, and Iterative Null-space Projection (INLP) removes a detectable framing direction without changing the SFT-to-final-RLVR gap. Together with preliminary 32B endpoint results (35.53% SFT to 22.37% RLVR final), these findings indicate that RLVR does not monotonically increase instrumental behavior in this setting: the rate rises early, then falls below the SFT baseline. The result is specific to this model family and benchmark, and judge agreement is moderate, so absolute rates should be read cautiously.

0.16

Harnessing Language Models as Optimization Modeling Agents

cs.AI ·2026-10-22 ·Yunxiang Luo, Xin Xu, Tianhao Chen, Zhixi Xiong, Peiqi Jiang +5 more
Abstract

Large language models (LLMs) can translate natural-language optimization problems into solver programs, but reliably producing correct formulations and implementations remains challenging. Existing approaches improve optimization modeling through agentic workflows or task-specific post-training. This raises a practical question: can a lightweight harness using a single model role achieve competitive performance without additional training? We introduce OPTICYCLE and OPTISTATE, two lightweight inference harnesses for optimization modeling, and evaluate them on nine benchmarks comprising 2,634 problems using 11 Qwen variants spanning 0.6B-27B parameters. Our results show that execution-guided revision consistently improves macro accuracy over direct generation for all evaluated models, with gains of up to 49.46 percentage points. Despite their simplicity, both harnesses outperform six adapted agentic workflows in same-model comparisons and four evaluated post-trained optimization models at comparable model sizes, without additional training. Explicit formulation state provides further but model-dependent improvements, suggesting that additional structure is not universally beneficial.

0.16

When Is a Memory Safe Enough to Act On? Action-Conditioned Verification for Long-Lived Agents

cs.MA ·2026-10-22 ·Yiqi Wang, Yifei Wang, Jiaqi Zhang, Zhangkai Wu, Yiqun Duan +2 more
Abstract · 22 Pages, 5 Figures, 12 Tables

Long-lived agents act on memory carried across sessions, and many of their actions are irreversible, so a stale memory can turn directly into harm. Existing work improves the memory itself --- retrieving, refreshing, or reconciling entries --- yet whether an entry must be re-checked before acting is not a property of the entry: the same remembered address can be used unchecked for a routine shipment but should be re-confirmed before a valuable one. We therefore study \emph{action-conditioned memory verification}: given a candidate action, decide which of its memory premises to re-verify and when to proceed or abstain, balancing execution risk, abstention, and verification cost. We formulate this as a sequential stopping problem, in which a verification is worth its cost only if it can change what the agent does next; under suitable verifier reliability and cost conditions, increasing action consequence can move the same memory through proceed, verify, and direct-abstention regimes. Building on this formulation, \textsc{AMV-Guard} is a practical verification gate that extracts the premises a candidate action rests on, calibrates beliefs about whether they are wrong, and selects checks by bounded value of information (VOI). On ToolSandbox, relative to a static policy given the same inputs, it lowers mean verification cost by 33.9\% while keeping 100\% safe-proceed recall and 0\% Gate harmful, and raises Proceed coverage from 38.83\% to 44.17\%; in a gate-core transfer to AppWorld, which carries over the stopping/VOI core but not the full structured pipeline, it lowers verification cost by 21.3\% at a small cost in safe-proceed recall and task success. These results suggest that verification effort should be set by what an agent is about to do, not only by how reliable a memory appears.

0.16

DIAL: Steerable Language Models with Distilled Instruction-Adherence Levels

cs.AI cs.LG ·2026-10-22 ·Thanh Q. Tran, John Russell Himawan, Arun Verma, Kiwan Wong, Wei Xiao +2 more
dataset
Abstract · This paper introduces DIAL, a framework for fine-grained, efficient, and scalable control of LLM instruction adherence

When a user says "Keep it short", how short should the LLM's response be? More generally, users and applications require different levels of instruction adherence, but prompting does not reliably provide fine-grained control over them. We therefore study "steerable language models" that provide fine-grained numerical control over the degree to which models follow general instructions specified in text. However, existing approaches do not jointly offer fine-grained control, efficiency, and scalability across diverse instructions. To address this gap, we introduce DIAL (Distilled Instruction-Adherence Levels) to achieve all three properties. For fine-grained control, we perform classifier-free guidance (CFG) with a frozen LLM to provide training signals for on-policy distillation of instruction adherence at different levels. For efficiency, we design a lightweight steering adapter that uses a classifier over the LLM's hidden states to compute a closed-form steering edit. The classifier adaptively sets the steering magnitude at each token based on the target adherence level, while accounting for the geometry of the LLM's hidden states to preserve response coherence. For scalability, the classifier is text-conditioned, allowing a shared adapter to support new instructions without per-instruction fitting, and training requires only instruction and request text. Evaluation across multiple LLM families shows that DIAL achieves a competitive balance between concept expression and response fluency, with concept scores increasing monotonically across adherence levels. We introduce SteerIF, an instruction-only dataset of 98, 625 concept, persona, verifiable, and descriptive instructions. When trained on SteerIF, DIAL provides fine-grained control over personas and, when combined with prompting, improves verifiable instruction following.

0.16

Rethinking Multi-View Fusion for Urban Representation Learning

cs.LG cs.AI ·2026-10-22 ·Sean Bin Yang, Zongyi Xu, Jilin Hu, Ying Sun, Kristian Torp +3 more
published
Abstract · Accepted by PVLDB 2027

Urban region representation learning encodes heterogeneous urban signals, such as mobility flows, points of interest, and land-use patterns, into unified embeddings for downstream tasks, including check-in prediction, crime forecasting, and service demand estimation. Existing approaches rely on globally shared fusion and joint optimization, implicitly assuming uniform modality importance and naturally balanced multi-objective training. However, urban data is inherently heterogeneous. Our empirical analysis reveals two structural issues: (i) modality utility varies markedly across regions, causing static fusion to overemphasize signals from certain modalities; and (ii) heterogeneous objectives exhibit persistent gradient imbalance, causing representational bias and unstable training. We present PRISM, a self-adjusting framework for adaptive multi-view learning. PRISM adapts sparse Mixture-of-Experts (MoE) routing to learn region-specific view allocations, while a Proportional-Integral-Derivative (PID)-inspired regulation mechanism mitigates gradient conflicts during optimization. Experiments on three large-scale urban datasets show that PRISM achieves more competitive performance than strong multi-view baselines, while providing more stable optimization behavior and more balanced region representations.

0.16

Contestable Agents for Data Governance: Recording Human Stewardship of AI-Generated Metadata Across Structured and Unstructured Systems

cs.AI cs.DBcs.HCcs.MA ·2026-10-22 ·Humza Akhtar, Brian Freeman
Abstract · 15 pages, 2 figures, 2 tables

Industrial organizations hold operational knowledge in structured records, such as telemetry, point lists, and transactions, and in unstructured sources, such as manuals, service notes, communications, and inspection reports. Governance programs often manage these sources separately. Classifying, linking, and documenting them depends on a small group of employees with deep domain knowledge. Agents built on large language models can propose metadata, cite documents, and flag quality issues, but unreviewed proposals can spread errors into downstream systems. A perfunctory approval leaves no record of what a reviewer checked. We present contestable governance: specialist agents produce scored, cited arguments for disputed metadata decisions; a steward can accept, reject, modify, or add arguments before commit; and a governance record retains the evidence, interventions, and outcome. A versioned domain glossary supplies model-specific terms and acronym definitions to the agents. We describe the workflow, a triage path that reserves full argumentation for uncertain proposals while preserving review before commit, the scoring and record models, and a gate that controls which verified decisions become precedent. Two HVAC use cases illustrate the pattern: point classification using Brick Schema class labels, supported by an open reference implementation, and an illustrative warranty failure-mode decision linking claim fields and telemetry to service notes and revisioned bulletins.

0.16

Lifetime-valid decisions by continually evolving AI analysts

cs.AI stat.MEcs.LGstat.ML ·2026-10-22 ·Xiaojing Sun, Yuhan Zeng, Xiao Wang
theory
Abstract

AI agents increasingly act as analysts whose decision can often be cast as hypothesis tests. Classical tests control the Type I error rate, the probability of rejecting a true null hypothesis, for fixed procedures. However, AI analysts introduce internal randomness and can evolve continuously over time. As AI analysts evolve, the risk of approving versions with Type I error rates above the target can accumulate, and adaptive use of evaluation feedback can lead to overfitting. We introduce continual validation, which evaluates each proposed AI analyst version before deployment using simulated data under the null hypothesis, with scenarios that can be randomized and expanded over time. We prove that, regardless of how updates are chosen or how many are proposed, the lifetime probability of ever deploying a version whose Type I error rate exceeds the allowed level remains bounded. This guarantees that every subsequent decision controls the Type I error rate at the target level across these scenarios. At the same time, updates with sufficiently low Type I error rates pass validation with high probability, allowing the analyst to continuously improve its performance without sacrificing rigor. In two examples involving evolving AI analysts for bioequivalence studies and product A/B tests, Type I error rates reached 0.098 and 0.183 without continual validation, compared with at most 0.045 and 0.044 with it, while retaining 82% and 96% of the power from optimally selected valid updates.

0.16

Faithful Latent Harm Monitors Work No Better Than the Models Own Behaviour at a Matched False Alarm Rate

cs.AI ·2026-10-22 ·Abhishek Mishra, Sagnik Chatterjee, Suyash Maniyar
Abstract

Large language models have been reported to encode the harmfulness of a request separately from their decision to refuse it, and latent safety monitors such as Latent Guard build on this report by classifying prompts as harmful from the internal activations of the model, with the aim of catching harmful intent even when the model complies. Whether these representations give a signal of harm that survives distribution shift and is independent of model behaviour is unclear. We test this on out-of-distribution prompts made by rewriting existing harmful and benign requests with surface rewording, persuasive framing and jailbreak-style wrappers. Across 15 instruction-tuned LLMs and 38 activation-based readers, we compare harm detection with that of TF-IDF and sentence-embedding classifiers, with independently trained safety classifiers, and with the refusal of each model. Activations are read at the last token of the user instruction, where the model has read the whole request, and at the last chat-template token before the reply, where the planned response can be represented. Because models disagree about which prompts to refuse, we can separate a signal of the harmfulness of the prompt from a signal of the refusal decision of the model. At the end of the instruction, the representation is largely independent of the refusal behaviour of each model but degrades under distribution shift, so that a TF-IDF classifier ranks harm better than the original Latent Guard in every model we test, and linear and MLP probes do about as well as sentence-embedding classifiers trained on the same examples. The representation just before generation detects more, catching 44 percent of harmful prompts at a matched false-alarm rate against 32 percent for the refusal of the model, but that gain follows the refusal decision itself, and part of it comes from refusals that string-matching evaluation misses.

0.15

The Helpful Agent Problem: When Emergent Cooperation Among LLM Agents Bypasses Intended Workflows

cs.AI cs.CLcs.MA ·2026-10-22 ·Moumita Choudhury, Amit Sarker, Shlomo Zilberstein
benchmark
Abstract

LLM agents are increasingly deployed in collaborative settings where they pursue separate tasks while sharing resources and operating under review or oversight processes. Such settings create safety risks not only when agents are malicious, but also when they are simply trying to help. Prior work has largely studied adversarial or explicitly collusive behavior; we show that benign agents pursuing separate objectives can spontaneously cooperate in ways that bypass intended workflows. We call this the \emph{helpful agent problem}. We study long-horizon shared-resource environments in which agents sometimes depend on resources owned by their peers and cross-owner changes are normally mediated by a reviewer. When review becomes unavailable, agents begin making changes on one another's behalf despite having no shared reward or explicit incentive to cooperate. Crucially, these actions remain within each agent's local permissions while circumventing the system's intended oversight structure. We introduce \textsc{FavorBench}, a controlled benchmark with measurable downstream side effects, and adapt \textsc{CooperBench} to test the phenomenon in realistic repositories. Across two benchmarks and four models, we find that out-of-workflow favors emerge frequently. Further, in \textsc{FavorBench}, favors from at least three models cause significant harm to parties outside the interaction. These results reveal a distinct multi-agent safety challenge: cooperation itself can create pathways around system-level safeguards.

0.15

SABLE: Scalable Affine Bayesian Latent-space Exploration

cs.LG cs.AIstat.ML ·2026-10-22 ·Erik Bodin, Lisa Maria Kreusser, Donney Fan, Henry Moss, Geoff Pleiss +1 more
Abstract

Complex objects such as engineering designs and proteins are often difficult to optimise because their design spaces exhibit high-dimensional structure and objective evaluations are expensive. Generative models provide continuous latent representations, but their latent spaces can have thousands to tens of thousands of dimensions, creating a regime where the latent dimension $D$ greatly exceeds the evaluation budget. We introduce SABLE, a Bayesian optimisation method for high-dimensional latent spaces with an isotropic distribution and limited budget. SABLE uses a Bayesian affine surrogate for prediction and acquisition, treating higher-order structure as surrogate discrepancy. After $N$ observations, the slope update has dimension at most $N-1$, yielding a decomposition of each query into an in-span component optimised by expected improvement (EI) in at most $N$ dimensions and an off-span component governed by a scalar norm. We show that the off-span norm implicitly selected by affine EI can lead to underexploration, and SABLE instead prescribes it using a finite-horizon schedule, without restricting search to a learned subspace. A recompression scheme reduces information to $\mathcal O(M)$ pseudo-observations, bounding refitting cost by $\mathcal O(M^2D+M^3)$ while keeping parameter inference, prediction, and acquisition decisions to tens of milliseconds. Across synthetic benchmarks and latent-space optimisation tasks, SABLE achieves high sample efficiency and reduces decision time by several orders of magnitude at large budgets.

0.15

Better Reasoning, Not Just Bigger Models: A Structure-Grounded Framework for Efficient and Reliable Inference

cs.AI cs.IR ·2026-10-22 ·Amirhossein Aghaei, Liangyan Li, Shiyi Yang, Shaohua Wang, Yan Wang
benchmark
Abstract

Scaling large language models has substantially improved linguistic fluency, yet reliable autonomous agents require more than fluent generation. In multi-turn inference, current agentic systems often produce semantically plausible but structurally ungrounded reasoning, especially under incomplete observations. We propose Structure-Grounded Uncertainty Reduction (SGUR), a training-free framework that complements token-level generation with active uncertainty reduction over structured hypothesis spaces. SGUR represents incomplete observations as a query graph and aligns them with candidate hypothesis graphs through a prior-weighted partial graph matching mechanism, where database-derived empirical priors assign greater importance to decision-relevant relations rather than treating all graph edges uniformly. The resulting inference state is iteratively refined by an active information-gain strategy that selects the next most informative node to query, systematically pruning the candidate space. We instantiate SGUR on a challenging PrimeKG-derived sequential medical diagnosis benchmark, demonstrating that structural grounding can partially substitute for model scale: starting from only two observed symptoms, SGUR achieves 92.5% Hit@3, outperforming the strongest Graph-RAG baseline by 9.4 percentage points while using approximately 5x fewer tokens. Notably, SGUR enables smaller backbone models to match or exceed the performance of models up to 10x larger, suggesting that structural grounding provides a more robust and efficient paradigm for scaling agent performance than relying exclusively on increasing model size.

0.15

OP-Print: Tracing Membership Imprints in On-Policy Self-Distillation

cs.LG cs.CR ·2026-10-22 ·Qi Li, Xinchao Wang
Abstract

On-policy distillation has been widely adopted for alleviating the distribution mismatch of off-policy distillation. Self-distillation further removes the need for an external teacher by using the student's own outputs, enhanced with extra context or demonstrations, as supervision. This teacher-free design makes on-policy self-distillation practical, yet its privacy risks remain largely unexplored. In particular, using two coupled views of the same training example may strengthen its memorization. Motivated by this, we present the first study of membership inference attacks against on-policy self-distillation. We investigate both gray-box and black-box threat models and propose a unified attack framework, named OP-Print. OP-Print builds on our finding that the supervision imprint in on-policy self-distillation is sparse and structured, concentrating on uncertainty-sensitive correction sites rather than across the entire rollout. Exploiting this property, we construct attack signals tailored to the two access levels: in the gray-box setting, OP-Print directly aggregates the privileged teacher-student residual correction along the same rollout; in the black-box setting, it measures restart-based output-basin stability as an observable surrogate for the hidden correction process. Across diverse models and distillation schemes, all baseline attacks remain essentially at random guessing even under gray-box access, whereas OP-Print is broadly validated to be effective in both gray-box and black-box settings.

0.15

Subgoal-aware Memory Graph for GUI Agent

cs.AI ·2026-10-22 ·Dezhao Luo, Daniel Beechey, Derek Yuen, Weilin Luo
Abstract

GUI agents aim to automate complex user tasks by interacting with mobile interfaces, but ef- fective decision-making in long-horizon set- tings requires reusing prior experience rather than relying solely on the environment ob- servations. Existing memory-based methods typically store experience either at the single- action level or as entire trajectories. Single- step actions are often too fine-grained to carry meaningful semantics, since the success of an individual action does not necessarily imply progress towards the overall goal. In contrast, whole trajectories are often too coarse-grained and difficult to reuse across different tasks. We propose a subgoal-aware memory graph that distils past trajectories into structured, reusable evidence at the level of subgoals. We define a subgoal as consecutive actions shar- ing a single intent, the granularity at which progress toward the task becomes meaning- ful. To reduce redundancy and support cross- task reuse, we organise subgoal experiences as nodes in a graph, where related memories can be merged, extended, and retrieved based on the environment observation. On SPA-Bench, our method improves success rate by 11.11% over the baseline.

0.15

Prediction Uncertainty Controlled Computer Model Calibration

cs.LG stat.MLstat.APstat.CO+1 ·2026-10-22 ·Ryan Heejoon Byun, Dayeon Yoon, Won Chang
benchmark
Abstract

Computer-model calibration is usually undertaken in service of prediction, yet neither Bayesian Kennedy-O'Hagan (KOH) calibration nor amortized simulation-based inference is built to control the predictive error of the calibrated simulator. When parameters are weakly identified or the data-model discrepancy is misspecified, the simulator-output bands induced by a nominal posterior level can be far from nominal. We propose a two-layer framework that targets predictive miscoverage directly. First, a conditional flow-matching model learns an amortized posterior for the calibration parameters, trained on simulator outputs perturbed with Gaussian-process discrepancies of varying structure. Second, conformal risk control arranges posterior draws for the target observation into nested candidate sets and selects the smallest one whose fixed-rank pointwise bands attain a prescribed time-averaged miscoverage on a synthetic calibration sample. Under exchangeability of calibration and target observations, the bands control expected miscoverage in finite samples. In a synthetic benchmark, the proposed bands are roughly 20% the width of KOH-MCMC bands with miscoverage near the 10% target. In calibrating an age-structured SEIR model to the 2010 Malawi measles outbreak, they are again sharper and near-nominal, and the weeks they fail to cover coincide with the asymmetric epidemic decline, exposing structural inadequacy of the mechanistic model.

0.15

Combinatorial Characterizations of the Learnability of Conditional Distributions

cs.LG stat.ML ·2026-10-22 ·Mihir Dhanakshirur, Ambuj Tewari
Abstract

We study the PAC-learnability of function classes that map inputs to probability distributions over a finite space. The learner observes i.i.d. input-label pairs, with each label sampled from the conditional distribution of an unknown target hypothesis in the class. We measure learning through excess risk under a bounded proper loss that satisfies two mild assumptions. We introduce the \textbf{fat$_γ$Natarajan dimension}, which generalizes both fat$_γ$ and Natarajan dimension, and prove that its finiteness at every positive scale is necessary and sufficient for learnability. We also introduce a \textbf{fat$_γ$Graph} dimension extending the classical \gra dimension and establish inequalities comparing the two dimensions at different scales, leading to an equivalent characterization of learnability. Finally, we show that a distribution-valued function class is learnable if and only if each of its coordinate classes is learnable in the probabilistic concept model.

0.15

Beyond Local Accuracy: Benchmarking the Rollout Reliability of PDE Foundation Models

cs.LG ·2026-10-22 ·Jaewan Park, Seid Koric, Giri Krishnan, Sudhakar Pamidighantam, Surresh Marru
Abstract

PDE foundation models enable data-efficient adaptation but are deployed autoregressively, so short-horizon accuracy need not imply reliable simulation. We ask whether the short-horizon advantage of a foundation-model recipe persists when its predictions are repeatedly fed back as inputs for a long horizon. We define error amplification as G=E(H_f)/E(H_s), which measures full-horizon error relative to short-horizon error. We evaluate seven models on seven two-dimensional time-dependent PDE datasets: scratch-trained FNO and DeepONet; fine-tuned DPOT, GPhyT, Poseidon, and Walrus; and frozen in-context VICON. We report VRMSE, relative L_2, and relative MAE. Six datasets provide paired short- and full-horizon scores, while CE-RM contributes short-rollout sweeps only. At least one fine-tuned foundation model outperforms both scratch baselines in 26 of 30 short-rollout dataset--budget cells under every metric. Nevertheless, full-horizon VRMSE exceeds short-horizon VRMSE in 140 of 144 paired trained configurations, with a median error amplification of $2.04$. The count below VRMSE one falls from 104 to 53, and 12 full rollouts exceed ten although none does at the short horizon. Walrus achieves the lowest full-horizon VRMSE in 16 of 24 cells, yet its median error amplification is $2.70$. Relative L_2 and relative MAE show the same qualitative pattern. These results show that favorable short-horizon rankings can coexist with substantial error amplification. PDE foundation-model benchmarks should therefore report horizon-resolved absolute errors, amplification statistics, and failure incidence rather than a single local forecast score.

0.14

ColaBench: Evaluating Memory-Guided Knowledge Acquisition in Personal Agents

cs.AI cs.CL ·2026-10-22 ·Wenjie Zhang, Zhichao Shi, Xuhui Jiang, Xiaojun Wu, Cehao Yang +2 more
benchmark
Abstract · 12 pages, 4 figures

Personal agents need to draw on long-term interaction history to retrieve and apply external knowledge suited to a user's current request. Even an answer grounded in relevant documents can be unsuitable when it overlooks the user's circumstances. This creates two connected challenges: using historical conditions to identify the knowledge needed for a new request, and applying that knowledge to the user's situation. Existing benchmarks advance memory recall, personalized retrieval, and memory-guided action, while the joint evaluation of these two challenges remains underexplored. We introduce ColaBench, a synthetic benchmark in health and daily living, with 349 tasks across 10 Profiles and a knowledge base of 13,033 documents. Its Evidence Graph connects personal conditions to their disclosures across public interaction sessions, records prior document exposure, and links answer requirements to personal and newly acquired document evidence. We also introduce MemKB, which uses state summaries to inform queries and rule application, and memory-document links to retrieve knowledge associated with the user's history. MemKB achieves 86.25% Accuracy with an average of 4.07 interaction rounds per task. Task-level analysis finds larger gains on tasks integrating information across multiple historical Facts; error analysis further identifies unsupported personal assumptions and rule misapplication as obstacles to using retrieved evidence. ColaBench provides traceable tasks and an evaluation framework for studying how memory guides knowledge acquisition and rule application.

0.14

Fine-Grained Analysis of Catastrophic Forgetting in Continual Learning via Algorithmic Stability

cs.LG ·2026-10-22 ·Guodong Zheng, Yan Sun, Peng Wang, Enneng Yang, Jiafei Lyu +3 more
Abstract

Understanding catastrophic forgetting, in which learning new tasks degrades performance on previously learned tasks, remains a fundamental challenge in continual learning (CL). Existing theoretical analyses often rely on specific model or task structures and offer limited insight into how iterative optimization affects forgetting. To address this gap, we develop an algorithmic stability framework for expected population forgetting under SGD without prescribing a particular model architecture. Specifically, we decompose forgetting into intra-task and cross-task generalization gaps and empirical forgetting. To bound the generalization gaps, we track how a sample perturbation affects updates within a task and propagates through subsequent tasks. We consider constant, intra-task decaying, and continuously decaying learning-rate schedules. Comparing the two decaying schedules allows us to characterize how restarting learning-rate decay at task boundaries affects the propagation of historical sample perturbations. For empirical forgetting, we derive bounds for all three schedules by retaining cumulative gradient alignment along the optimization trajectory, without requiring convexity. Combining these results yields upper bounds on expected forgetting that capture the effects of training set size, the number of updates per task, the total number of tasks, learning-rate schedules, and cumulative gradient alignment. Finally, we conduct experiments on Split CIFAR-100 and Split MiniImageNet to examine the empirical implications of our theoretical findings.

0.14

Dimension-free regimes for unadjusted Langevin in ridge-separable models: A subspace robust Wasserstein analysis

cs.LG stat.MLmath.PRmath.ST+1 ·2026-10-22 ·Yuansi Chen, Francesco Pedrotti, Peter A. Whalley
theory
Abstract · 30 pages, 4 figures

The Unadjusted Langevin Algorithm (ULA) is widely used for high-dimensional Bayesian inference. Classical analyses in Wasserstein-2 distance predict an asymptotic bias and a mixing time that grow with dimension, but this metric is stronger than necessary for many statistical tasks. We instead analyze the bias in subspace robust Wasserstein distance, which controls the mean and arbitrary low-dimensional marginals and is therefore well suited to observables typically estimated in practice. For ridge-separable potentials formed with i.i.d. data, we prove high-probability bounds showing a bell-shaped mixing profile as the number of data points $N$ varies relative to the dimension $d$. In the overparameterized regime, the effective dimension is the rank $N$ of the data-dependent perturbation, leading to complexity depending on $N$ rather than $d$. In the large-sample regime, Gaussian-perturbation estimates yield dimension-free complexity once $N$ is sufficiently large. In contrast, near the proportional regime $\asymp d$, we prove that dimension-free guarantees are impossible in general.

0.14

Scaling Transformer Energy-based Models For High Quality Image Synthesis

cs.LG cs.CV ·2026-10-22 ·Jungsoo Lee, Alexi Gladstone, Yilun Du
Abstract

Energy-Based Models (EBMs) are a flexible class of generative models that decouple probabilistic modeling from the generation process. This decoupling directly enables compositional generation and inference-time scaling, where sample quality can be refined simply by increasing sampling steps. However, severe challenges in high-dimensional sampling and training instabilities with deep architectures have restricted EBMs to low-dimensional regimes. To resolve these bottlenecks, we introduce the Transformer Energy-Based Model (TEM), a scalable transformer energy function operating entirely within a latent space. On top of this backbone, we extract energies at multiple layers and sum them, composing complementary constraints from local texture to global semantics. Furthermore, two components make this high-quality at scale: Positive Representation Alignment (PRA), which supervises TEM's intermediate representations with a vision foundation model, and advanced sampling strategies that improve sample quality at inference. TEM-XL achieves an FID-50K of 3.82 on class-conditional ImageNet 256$\times$256 without classifier-free guidance, showing that a contrastive EBM can generate high-fidelity images. On ImageNet 512$\times$512, TEM achieves an FID-50K of 8.24, demonstrating that the approach scales to higher resolution. We further validate the core promises of EBMs: compositional generation, zero-shot inpainting, and out-of-distribution detection.

0.14

Continuous-depth Deep Gaussian Processes

cs.LG stat.ML ·2026-10-22 ·Ying Li, Sanyou Wu, Zi Yang, Qiaochu Xu, Yuhao Liu +4 more
benchmarkposition
Abstract · 22 pages, 3 figures

Deep Gaussian processes (DGPs) are appealing Bayesian regression models, but their standard discrete-depth parameterization often becomes harder to optimize as depth grows. We argue that depth is better treated as a latent continuous evolution than as a long discrete stack. We therefore propose a continuous-depth formulation in which the latent representation follows a stochastic differential equation with a depth-indexed drift family equipped with a GP-based prior. This reframes the core design choice as the prior placed on that family, rather than adopting a purely neural continuous-depth parameterization as in neural ODE or neural SDE models. We study two instantiations of this idea: CDGP, a direct GP prior over the state--depth drift, and FlowDGP, a flow-evolved prior that captures richer depth dependence at lower training cost. To motivate the redesign, we analyze how the single-sample DGP objective can become increasingly sensitive under repeated layer composition, while residual DGP is a useful first repair that still remains a finite discrete construction. We also give pathwise sensitivity analysis for the continuous-depth formulation. Experiments on synthetic, benchmark, and large-scale regression tasks show that residual DGP improves over plain DGP, while CDGP and FlowDGP provide a stronger overall story across the regimes we study.

0.14

AgentDP: Towards Automated Model Deployment for Heterogeneous Embedded AI Applications

cs.AI ·2026-10-22 ·Leming Shen, Qiang Yang, Xinyu Huang, Zijing Ma, Yuanqing Zheng
Abstract

This paper presents AgentDP, which aims to deploy pre-trained AI models on heterogeneous embedded devices by fully automating model optimization and adaptation (O&A) with Large Language Model (LLM) agents. AgentDP not only significantly reduces human effort via automation but also enhances model inference efficacy by integrating LLMs with device-specific domain knowledge. To deliver a robust automation framework, AgentDP develops novel solutions to address three key technical challenges: 1) To help users elicit clear requirements, we propose a user requirement standardization module; 2) To mitigate the execution environment mismatch between AI models and heterogeneous embedded devices, we design an operator-aware model partition paradigm that enables different parts of the model to be efficiently executed on distinct inference engines. 3) To avoid re-searching optimization hyperparameters for new tasks, we propose a configuration reuse module that heuristically retrieves previous hyperparameter configurations for efficient reuse. Extensive experiments demonstrate that by fully automating the entire O&A process, AgentDP reduces entire model adaptation time by up to 80% compared with manual O&A while achieving higher task accuracy (more than 13% on average) than the baselines. Additionally, a comprehensive user study also exhibits the robust generalizability of AgentDP in adapting to different user requirements for heterogeneous embedded devices.

0.14

Diffusion Language Models for Perplexity-Aware Black-Box Attacks

cs.LG cs.CLstat.ML ·2026-10-22 ·Ali Al-Lawati, Elsayed Eshra, Dongwon Lee, Suhang Wang
Abstract

\emph{Continuous-space} attacks find prompts that trigger malicious behaviors, such as overthinking, in black-box LLMs by sampling or optimizing over a low-dimensional Gaussian latent space and decoding it to text through a \emph{random mapping} and nearest-token lookup. Because the random mapping is not learned from language, its prompts can be highly unnatural, and as such easily detectable through input guardrails, and the risk estimates they yield are not grounded in natural prompts. In this paper, we demonstrate that the same continuous latent-space framework can be coupled directly to a learned generative distribution over language, a continuous diffusion language model (dLLM), which removes the artificial random mapping while preserving the standard-Gaussian latent formulation and the downstream sampling or optimization procedure. We evaluate this approach under two settings: sampling-based estimation using \emph{RareTrap} and optimization-based search using \emph{ThinkTrap}. We find that character-level perplexity drops from $10^3$--$10^7$ to 3--5 in RareTrap and from 20--28 to 4.3--5.4 in ThinkTrap. We also find that model risk estimates can differ by an order of magnitude across the two mappings.

0.14

End-to-End Differentiable Diffusion Conditioning for Physics-Informed Optimization

cs.LG cs.AI ·2026-10-22 ·Ricardo Luna Gutierrez, Vineet Gundecha, Rahman Ejaz, Varchas Gopalaswamy, Riccardo Betti +3 more
Abstract · Initial version submitted to the NeurIPS 2026 main conference pending publication

Optimizing high-dimensional physical systems under expensive evaluations and complex constraints is a central challenge in science and engineering. In offline model-based optimization (MBO), methods must improve designs using only a fixed dataset and no test-time access to the true objective. However, existing methods typically suffer from conservatism, unconstrained surrogate exploitation, or function as static inference-time samplers that cannot refine candidates against precise physical objectives. In this paper, we propose Differentiable Diffusion Conditioning (D2C) for offline MBO, which leverages the fact that deterministic sampling makes the reverse denoising chain differentiable with respect to the conditioning signal. D2C turns a frozen conditional diffusion model into an end-to-end gradient-based optimizer by directly optimizing conditioning variables at test time via gradients from frozen surrogate objectives and physics-informed penalties backpropagated through the reverse chain. We implement this idea in a composed proposer-evaluator framework and evaluate it on laser pulse optimization for inertial confinement fusion, sustainable data-center workload scheduling, and four Design-Bench tasks. Our method achieves the strongest results on the two real-world tasks and the best average rank across four Design-Bench tasks.

0.14

XALIGN: A Human-centric Visual Analytics Approach to Understanding AI Explanations

cs.LG ·2026-10-22 ·Bahavathy Kathirgamanathan, Gennady Andrienko, Natalia Andrienko, Tatiana von Landesberger
Abstract

Explainable AI (XAI) methods are usually evaluated for technical properties such as faithfulness and robustness. In high-stakes application domains, experts also need to check model predictions against established knowledge. Different explanation methods often disagree, and this disagreement is hard to see when each method is inspected separately. We propose XALIGN, a visual analytics workflow and interface that shows rule-based explanations and several feature attribution methods (TreeSHAP, KernelSHAP, and LIME) side by side for the same Random Forest model. Agreements and contradictions between the methods become visible signals that users can explore and compare with domain knowledge. We apply XALIGN in two case studies: the German Credit dataset and a meteorological dataset, in which events derived from ICON model output and Cloudnet observations are classified. In both, the views reveal discrepancies between the rule-based explanation and the attribution methods that a single method would not show. In an expert evaluation with three specialists, participants found XALIGN useful for identifying influential features and for avoiding over-trust in predictions. Linking rules to attributions was harder for the participant without an XAI background.

0.13

Drift Says When, Budget Says How Much: Rate-Controlled KV Re-selection for Long-Form Multimodal Generation

cs.LG ·2026-10-22 ·Bin Hu, Yingbo Xiao
Abstract · 17 pages, 7 figures, 13 tables

KV cache compression for multimodal LLMs usually selects the visual tokens to keep once, at prefill, and reuses that selection for the whole generation. We show that this one-shot selection is severely limited on long-form generation: on detailed image description, even an oracle that selects with the attention of the uncompressed answer reaches only 0.44-0.52 ROUGE-L agreement with the uncompressed output, because the evidence the model reads moves as the description unfolds. Re-selecting whenever attention drifts out of the kept set is the natural remedy, but we find that a fixed drift threshold fails systematically: drift magnitudes are not comparable across samples, so low-drift samples are almost never refreshed although they benefit the most. We propose Rate-Controlled Refresh (RCR), which lets drift decide when and where to re-select while a per-sample controller regulates how often, combined with a low-rank drift detector, candidate-restricted re-selection, and a decision that stays on the GPU. With one setting for all models and tasks, RCR outperforms the per-step retrieval method InfiniGen by 1.6-3.8 ROUGE-L points on five of six model-task settings with no significant difference on the sixth, while staying within 1.7 points of exact per-step re-selection, with comparable estimated data movement per step and a 4x smaller GPU-resident detector. On DetailCaps, drift-chosen timing beats random timing at the same refresh rate and periodic timing even at a higher rate, by 2.0-5.0 points. On pre-registered video description, where drift is homogeneous across samples, RCR is on par with InfiniGen on Qwen2-VL and exceeds it on Qwen3-VL but no longer beats a fixed threshold at the nearest tested rates, consistent with the diagnosis; object-level evaluation shows that RCR achieves higher object F1 against the uncompressed output than InfiniGen on two of three models.

0.13

Harmful Responses Are Jailbreak Templates: Transferring Autoregressive Jailbreaks to Diffusion Language Models

cs.LG ·2026-10-22 ·Jinwei Gan, Zifeng Cheng, Zhiwei Jiang, Ran Sun, Shiping Ge +1 more
Abstract

Diffusion large language models (dLLMs) enable flexible masked generation through bidirectional context modeling, but this capability also introduces a distinct jailbreak attack surface. We find that a harmful response to a given query naturally serves as an effective jailbreak template: simply masking a subset of its tokens can induce a dLLM to reconstruct harmful content. Based on this observation, we propose RMT-Jailbreak, a general framework that transfers existing autoregressive (AR) jailbreak attacks to dLLMs through their induced harmful responses. RMT-Jailbreak converts an AR-generated harmful response into a high-mask-ratio template using uniform token retention, and further applies prompt-based safety filtering to mask any remaining visible harmful content. The refined template is then provided to the target dLLM for bidirectional reconstruction. Extensive experiments across multiple AR jailbreak methods and target dLLMs demonstrate the effectiveness and generality of RMT-Jailbreak, revealing strong cross-paradigm transferability of jailbreak attacks.

0.13

Beyond Causal Masks: Auditing and Localizing Hidden Future-Token Leakage in Adaptive Sequence Models

cs.AI ·2026-10-22 ·Huiting Liu, Jinsong Wu
Abstract

Hidden future-token leakage can create false architectural gains: an autoregressive model may appear substantially better under teacher-forced evaluation while its free-running generation deteriorates. We investigate this failure in an adaptive residual-routing decoder whose sequence-level routing summary exposes prefix positions to future tokens despite a correct causal attention mask. An anomalous 24-layer checkpoint reduces validation loss from the baseline's 3.5958 to 1.4259, yet it fails a prefix-invariance audit and has substantially higher reference-model perplexity during generation. Layer-wise diagnostics and implementation inspection trace the violation to a full-sequence summary broadcast through the dynamic routing-query path. Two independently trained 12-layer causal repairs eliminate detectable leakage and return validation loss and reference perplexity to the control scale, removing the large apparent advantage. As diagnostic tools, we use the established prefix-invariance contract, maximum prefix-output discrepancy (LeakScore), complementary behavioral metrics, and layer-wise profiles. Supporting experiments cover six synthetic implementation pairs spanning five module families, clean pretrained controls, controlled fault injection, and cache consistency across numerical precisions. These results illustrate how hidden future-token access can produce misleading claims of architectural progress and why lower teacher-forced loss should be interpreted together with whole-model causal checks and free-running evaluation.

0.13

Knowing Whom to Trust is not Enough: The Trust--Agreement Gap in Repeated Multi-Party LLM Negotiation

cs.AI cs.CLcs.MA ·2026-10-22 ·Amit Sarker, Moumita Choudhury, Saaduddin Mahmud, Shlomo Zilberstein
Abstract

Large language model (LLM) agents increasingly participate in negotiations on behalf of individuals and organizations, yet they remain vulnerable to manipulation, particularly in repeated multi-party interactions. Classical multi-agent systems address this challenge through explicit trust and reputation models, whereas existing work on LLM negotiation largely treats trust as an emergent phenomenon. We investigate whether LLM agents can form useful trust estimates from prior interactions and leverage them to improve future negotiation outcomes. We introduce a repeated, multi-party, multi-issue negotiation framework in which recurring agents participate in a sequence of negotiation games while one participant covertly attempts to prevent agreement. After each game, agents assess their counterparts and maintain pairwise trust estimates that carry over to subsequent negotiations. We compare trust-aware and trust-agnostic agents and analyze how explicit trust influences negotiation outcomes. Our results reveal a \emph{trust--agreement gap}: trust-aware agents can identify the adversary within a few games, but accurate trust estimates do not consistently translate into better agreements. Trust increases agreement rates by up to 23 percentage points, but only when agents are explicitly guided to use it in their decisions. Explicit trust can therefore improve repeated LLM negotiations, but its value depends critically on how trust is assessed and incorporated into agents' reasoning.

0.13

Same Luck, Different Choice: Twin-World Credit Assignment for Agentic Reinforcement Learning

cs.LG ·2026-10-22 ·Zeropy Surio, Tobias Engel, Mei-Lin Chou, Anika Raman, Samuel Okafor
Abstract · arXiv admin note: This submission has been withdrawn by arXiv administrators due to incorrect author affiliation listing

Agentic reinforcement learning rewards a complete episode, yet the policy needs credit for every decision inside it. Counterfactual branching estimates this credit by re-running the episode after an alternative decision, but the branch draws its sampling noise independently of the factual episode, so the outcome difference mixes the effect of the decision with the luck of everything that follows. We observe that in agentic search the retriever, the corpus, and the grader are deterministic, so an episode is a deterministic function of the policy's decisions and its sampling noise. Twin worlds are therefore realizable: the same noise, a different decision. We introduce TWICE (TWIn-world Credit Estimation), which keys the Gumbel noise of every token by turn and line, rolls out counterfactual twins that inherit the factual episode's downstream noise, and turns their paired outcome differences into turn-level advantages for policy-gradient updates on actor-sampled trajectories; only the actor is deployed. Twin credit is unbiased for any coupling quality, its variance is that of an independent counterfactual scaled by $1-ρ$, where $ρ$ is the twins' outcome correlation, and under coupling the number of rollouts needed to resolve an advantage of size $Δ$ falls from $Θ(Δ^{-2})$ to $Θ(Δ^{-1})$. A square-root rule spends the twin budget where outcomes are predicted to disagree per unit cost. On seven question-answering benchmarks, TWICE improves the average exact match of GRPO by 5.3 and 4.5 points on Qwen2.5-3B and 7B, outperforms tree-search and counterfactual-branching baselines, and uses 68% fewer auxiliary tokens than VinePPO.

0.13

Causal Chain of Thought: Mitigating Attention Leakage via Context Isolation

cs.LG cs.AI ·2026-10-22 ·Gabriele Dominici, Teague McMillan, Fazl Barez, Marc Langheinrich, Martin Gjoreski
Abstract

Chain-of-thought (CoT) prompting elicits step-by-step rationales that appear to explain model predictions, but there is no guarantee these steps causally drive the final answer. In standard autoregressive generation, dense attention allows each reasoning step to attend to the full context, enabling the model to derive answers through direct lookup while generating plausible-sounding but post-hoc justifications. We introduce CausalCoT, an inference framework that enforces causal transparency by decomposing problems into a directed acyclic graph of atomic sub-questions, evaluating each node in an isolated context containing only its direct causal parents. This Markovian structure removes attention pathways to non-parent content, making downstream predictions sensitive to local reasoning rather than global context leakage. Across a diverse suite of reasoning benchmarks, CausalCoT preserves competitive predictive performance while improving explanation faithfulness, increasing causal sensitivity of intermediate reasoning steps by an average of 12 and up to 21 percentage points over the strongest baseline. The structured dependency graph additionally enables test-time interventions in which local corrections propagate predictably through downstream reasoning.

0.12

RIPPLE: Recovery Internalization through Process-Level Policy Learning and Effect-Aware Credit for Long-Horizon Agentic Coding

cs.AI ·2026-10-22 ·Jingxuan Xu, Zihang Xu, Yuechen Luo, Ziqi Miao, Wenqiang Zhu +7 more
Abstract

Long-horizon coding agents can observe a failure and generate a plausible critique yet continue along the same incorrect trajectory. We call this the Critique-to-Action Handoff Gap. We introduce RIPPLE, a recovery-oriented training framework that internalizes process-level guidance into a single self-recovering policy. A prefix-only teacher provides evidence-grounded recovery supervision without access to future actions or terminal outcomes. After lightweight distillation, reinforcement learning assigns stage-specific credit to reflection triggering, diagnosis, and corrective action. Local validity signals gate trajectory advantages so that intermediate recovery decisions do not indiscriminately inherit terminal outcomes. Across four repository-level software-engineering benchmarks, RIPPLE improves the base policy by an arithmetic mean of 4.8 score points, with no auxiliary critic at deployment. Controlled ablations isolate the benefits of local, stage-aligned credit, and matched-prefix interventions show that critique semantics affect subsequent correction. Recovery gains concentrate where the prefix contains sufficient diagnostic evidence and the repository state remains reversible within the remaining budget.

0.12

Agentic Knowledge-Model Co-Evolution via Consistency Regularization

cs.AI ·2026-10-22 ·Sirui Chen, Zhining Liu, Chao Zhang, Jingrui He, Zheng Li
Abstract

LLM agents can evolve by learning from accumulated experience. Existing approaches extract reusable external knowledge, such as skills and memories, from experience to guide subsequent task solving, with some further internalizing this knowledge through model training. However, different ways of analyzing and transforming the same experience can yield knowledge with substantially different utility, even when the resulting guidance appears reasonable. Inspired by consistency regularization, we identify consistent utility gains across different ways of analyzing the same experience, rather than relying on any single way of analyzing it. We introduce SPLICE, a framework for consistency-regularized knowledge-model co-evolution. It constructs evolution views by perturbing how the same experience is analyzed, producing different knowledge candidates. Knowledge updates are then guided by common gain, which accounts for utility gains and their consistency across views. During model evolution, self-distillation internalizes the evolved knowledge alongside reinforcement learning, using knowledge associated with consistent utility degradation as a contrastive reference. The two processes interleave, with model updates driving further knowledge evolution. Experiments across diverse domains show that SPLICE outperforms knowledge evolution and model training baselines by up to 7.1 and 4.6 points in average accuracy, while its evolved knowledge reduces output tokens per correct solution by up to 44% and transfers across model sizes and families.

0.11

Constraint Reasoning and Large Language Models Join Forces: Providing Contrastive Explanations for Real-World Optimisation-Based Applications

cs.AI ·2026-10-22 ·Roger X. Lera-Leri, Filippo Bistaffa, Athina Georgara, Juan A. Rodríguez-Aguilar
Abstract · Currently under review at Applied Artificial Intelligence

We focus on optimisation-based Decision Support Systems in real-world applications. In this context, we aim to provide contrastive explanations that answer users' questions about recommendations based on the solution of an optimisation problem, specifically a Mixed-Integer Program (MIP). Addressing this endeavour is crucial to ensure human oversight of the decisions implemented, as required by international guidelines on trustworthy and responsible automated decision-making systems. We propose a novel methodology that combines constraint reasoning techniques and Large Language Models (LLMs) to generate contrastive explanations for optimisation problems encoded as MIPs. Although LLMs have shown impressive capabilities in Natural Language processing, they can produce incorrect answers due to reasoning limitations. To overcome these limitations, we leverage constraint reasoning techniques to distil the information necessary to produce an explanation, which is then passed as input to an LLM as the so-called graph of reasons. Such a graph captures the semantics and the relationships among constraints of the optimisation problem. Results on real-world scheduling domains show that incorporating constraint reasoning improves explanations generated by the LLM. In particular, using the graph of reasons enables the LLM to produce natural-language explanations that include more relevant reasons while reducing errors compared to explanations generated without this structure.

0.10

Inside-JEPA: From Prediction to Relational Grounding in Latent World Models

cs.LG cs.CVcs.RO ·2026-10-22 ·Hossein Resani, Javen Qinfeng Shi
Abstract

Joint-embedding predictive architectures (JEPAs) learn world models in latent space, enabling efficient planning but leaving their states opaque. We introduce Inside-JEPA, which places relational channels for ordered entity pairs inside the JEPA latent state and shows that an inverse-dynamics loss can ground these channels without relation labels when its head reads them, either alone or as part of the full state. On Something-Something V2, latent prediction alone performs no better than a random relation head, while inverse-dynamics supervision that recovers actions from channel changes improves relational AUROC by .065-.103 across ten settings with annotated boxes, by .081 on RH20T robot episodes with held-out tasks, and by .050-.061 across five relations using off-the-shelf detection and tracking. An inverse head on entity latents decodes actions more accurately but barely improves relational grounding; shuffled labels, two label-free objectives, and action labels on only 5% of the pairs also fail to ground the channels. Given an action, predicted relational changes approach the quality of observed future relations (AUROC .599 vs .621). For one-step relational goals, a supervised readout of the 16 channels ranks the correct action among the top five of 65 in 43% of cases, compared with 24% for prediction-only latents and 17% for a frequency prior.

0.10

Trion-1: Unifying Language, Protein Sequence, and Structure with Diffusion Language Models

cs.AI ·2026-10-22 ·Tianyu Lu, Kaicheng Yang, Jiangtao Feng, Keyue Qiu, Yawen Ouyang +5 more
Abstract

We envision a general-purpose protein foundation model as a unified, interactive model of sequence, structure, and function, capable of understanding, generation, and cross-modal translation while retaining the instruction-following and reasoning capabilities of modern LLMs. Toward this vision, we introduce Trion-1, the first native, unified, multimodal diffusion language model (dLLM) for proteins. Trion-1 is built on two key design choices: (1) a unified discrete token space that jointly represents sequence, structure, and natural language; (2) a dLLM backbone that replaces strict left-to-right factorization with bidirectional or semi-bidirectional token interactions, enabling global residue communication and long-range contact modeling that are naturally suited to structure modeling. This formulation unifies six core capabilities: folding, inverse folding, sequence understanding, structure understanding, sequence design, and structure design, while retaining the native language capabilities of general-purpose LLMs. On established benchmarks, Trion-1 achieves broadly competitive performance. In folding, Trion-1 achieves a TM-score above 0.8 with a single sample under strict homology control, matching or exceeding the existing tri-modal model using 200 samples. For sequence understanding, Trion-1 performs on par with top-tier LLMs such as Claude Opus 4.7 (62.37%). These results support well-aligned sequence-structure-language representations and strong cross-modal generative capabilities.

0.10

Measuring the Reasoning-Action Coupling in Driving Vision-Language-Action Models through Counterfactuals

cs.AI ·2026-10-22 ·Gabriel Manalu, Saad Wazir, Thao Do, Seongah Kim, Hyeontaek Hwang +2 more
Abstract · 11 pages, 4 figures, 3 tables

Driving vision-language-action (VLA) models bring an appealing form of explainability to a safety-critical domain, since a system that can state the reasons for its maneuvers offers the transparency that deployment and public trust in autonomous driving demand. These models read the world into language before they act on it, describing the scene and reasoning over it in text that ranges from chain-of-thought to multi-stage question answering on the way to a trajectory. That framing quietly assumes two separable properties, namely that the trajectory follows the generated reasoning and that the generated reasoning tracks the causes that actually produce the trajectory. To thoroughly investigate these properties, we apply causal, counterfactual interventions to four driving VLAs across different benchmarks. First, we perturb the models' own reasoning and show that altering its meaning produces small trajectory changes relative to blanking the same models' camera inputs, a pattern that persists across benchmarks and model architectures. We then construct scene-level counterfactuals through photorealistic removal of key objects named in the models' reasoning and analyze whether the generated reasoning and trajectories track these changes. Overall, our findings indicate that the reasoning these models present to account for their resulting driving behavior is weakly coupled to that behavior itself, pointing to limits on the explainability these systems are intended to provide.

0.09

When and How Does Muon Precondition? The Role of Structured Gradient Noise

cs.LG ·2026-10-22 ·Xiaoran Chen, Binghui Li, Liwei Wang, Pinyan Lu
Abstract · 62 pages, 11 figures

Muon has become a popular optimizer for large language model training, yet its theoretical foundations remain poorly understood. In this work, we study how gradient noise geometry shapes Muon's preconditioning benefits. We first show, theoretically and empirically, that under the classical isotropic noise assumption, Muon offers no acceleration beyond a rescaling of the learning rate. This motivates us to investigate structured noise arising from mini-batch sampling in associative memory with heterogeneous task frequencies. We establish an exact balancing threshold for two tasks and extend the analysis to multiple tasks, revealing how task coverage and gradient reliability jointly shape learning. Our analysis shows that Muon can soften the dependence on task frequency from linear to square-root, helping balance the learning of frequent and rare tasks. We further identify rotation-induced negative transfer in SignSGD that Muon avoids, and explain how momentum reduces sampling noise while tracking changing gradients. Controlled numerical experiments validate these mechanisms, and a preliminary LLM pretraining study illustrates the practical promise of fast-slow momentum.

0.09

One Pretrained Decision Model Is Enough for Training-Free Generalist Graph Anomaly Detection

cs.LG ·2026-10-22 ·Xinyu Zhao, Qingyun Sun, Ziwei Zhang, Tianchen Zhu, Daqing Li +1 more
Abstract

Graph anomaly detection (GAD) aims to identify irregular entities in graph-structured data, yet heterogeneous anomaly patterns and distributional differences across graphs challenge generalizable detection. Existing training-free approaches avoid task-specific optimization but primarily rely on predefined numerical scoring rules, leaving the semantic interpretation of heterogeneous anomaly evidence largely unexplored. In this paper, we propose Jev-GAD, a training-free generalist graph anomaly detection framework that bridges heterogeneous numerical evidence and pretrained decision mechanisms. To address heterogeneous anomaly patterns, we construct complementary structural, attribute, and neighborhood evidence and transform heterogeneous statistics into graph-relative anomaly percentiles. To move beyond predefined numerical scoring rules, we leverage a single frozen pretrained language model for evidence-conditioned anomaly assessment. To bridge the semantic gap between numerical evidence and pretrained models, we design a semantic Graph-to-State (G2S) transformation that organizes graph context and semantically labeled node evidence into structured decision states for non-generative option-logit readout. The resulting decision scores are integrated with four complementary statistical scores through equal-weight rank-based fusion, without GAD-specific gradient-based optimization. Extensive experiments on seven graph anomaly detection benchmarks spanning citation, social, and co-review networks demonstrate the effectiveness of Jev-GAD, achieving average AUROC and AUPRC scores of 89.18\% and 52.40\%, respectively. Further analyses validate the contributions of pretrained decision readout and semantic evidence grounding.

0.08

Reverse-KL Policy Distillation: Statistical Guarantees and Limits

cs.LG ·2026-10-22 ·Haolin Liu, Chen-Yu Wei, Julian Zimmert
Abstract

Reverse Kullback-Leibler (KL) distillation has shown empirical promise for training language models, but its statistical guarantees and limits remain poorly understood. In particular, the effects of two design choices remain unclear: Does access to the teacher's full action distribution provide an advantage over observing only the probability it assigns to the sampled action? Must training trajectories come from the student (online distillation), or can teacher-generated trajectories (offline distillation) suffice? Our work takes a first step toward understanding the theory underlying these questions. We study the sample complexity of online reverse-KL distillation under two feedback models: the teacher reveals either its full action distribution or only the probability of the student-generated action. We establish minimax bounds for the former and show that substantially more data may be required for the latter. In contrast, offline reverse-KL distillation can attain the same minimax rate whether the teacher reveals its full distribution or only its probability of the teacher-generated action. The key distinction is that teacher-generated actions themselves reveal the teacher's preferences, while queries at student-selected actions can leave those preferences unobserved. This highlights an additional statistical challenge for online distillation compared to offline distillation. These statistical guarantees do not follow simply from optimizing standard distillation objectives: our lower bounds show that common online objectives can fail when candidates are evaluated only on past student-generated trajectories. We develop trajectory-loss and importance-ratio truncation techniques together with policy-space learning to achieve optimal online distillation guarantees.

0.08

Beyond On-Policy vs. Off-Policy: At What Level Should LLM Agent Distillation Follow the Student?

cs.AI ·2026-10-22 ·Ya Gao, Zhengtao Zou, Pekka Marttinen
position
Abstract · Under review

Distilling interactive LLM agents requires choosing between teacher-generated demonstrations and student-generated experience, corresponding broadly to off-policy and on-policy distillation. While both have proven effective, we argue that this distinction is inherently hierarchical: the policy that induces an interaction history need not be the same as the policy that generates the token prefix of the current action. We introduce a hierarchical agent distillation framework that formalizes these two levels of on-policiness and enables explicit control over their pairing, mixture proportions, and training schedule. We study this problem through self-distillation, where a frozen teacher has access to task guidance unavailable to the student, and conduct a three-stage controlled study across three models and four interactive environments. We find that student-induced histories paired with teacher-generated action prefixes generally provide the strongest standalone training signal, whereas following the student at both levels generally performs worst. Nevertheless, fully on-policy data becomes beneficial when incorporated into a mixture with other configurations. Moreover, mixtures with identical teacher/student proportions at each level can yield substantially different performance, showing that how the two levels are paired matters beyond the overall amount of on-policy data. Finally, front-loading fully teacher-generated contexts and delaying fully on-policy contexts improves over a static mixture. These findings provide practical principles for deciding where, how much, and when agent distillation should follow the student.

0.08

Gandalf: Flash Shall Not Die! Increasing Flash Storage Lifetime for KV Cache Offloading in LLM Serving

cs.LG ·2026-10-22 ·Maharani A. P. Irawan, Kuntai Du, Jiayi Yao, Dongjoo Seo, Yuyang Huang +2 more
Abstract

KV caching reduces the serving cost of large language models (LLMs), but the growing KV cache footprint is increasingly exceeds GPU and host-memory capacity. Therefore, people use SSD to expand the KV cache storage capacity with high I/O bandwidth. However, in production LLM workload, such as multi-round conversation and LLM agents, the total amount of KV cache writes significantly exceeds SSD capacity, shortening SSD lifetime from years to months. To improve SSD lifetime, a natural solution is to reduce the KV cache writes by gating the KV cache writes via historical access patterns. However, that significantly lowers the KV cache hit rate (and increase the LLM serving cost) since KV caches from different attention types might have identical historical access pattern but different future access patterns. Our insight is that, to reduce KV cache writes without hurting KV cache hit rate, one need to not only consider the access pattern, but also application-level information such as the attention type of KV cache. Leveraging this observation, we propose Gandalf , the first SSD KV cache offloading system with SSD life time awareness. Gandalf prioritize the admission of full-attention KV cache into SSD over the cache generated by other types of attention, which significantly increases SSD lifetime without sacrificing cache hit rate. Our evaluation shows that Gandalf extends the SSD lifetime by 7.3X on average, while having negligible impact on cache hit rate across multiple workload traces and wide range of hardware capacity. Gandalf design can be extended to other flash-based storage, such as High Bandwidth Flash (HBF).

0.07

Can Entanglement-Induced Feature-Map Designs Improve Projected Quantum Kernels?

cs.AI quant-phcs.LG ·2026-10-22 ·Ravi Kumar Jha, Hsuan-Yin Lin, Eirik Rosnes
Abstract

Projected quantum kernels (PQKs) have emerged as a notable approach to circumvent the hardness of machine learning tasks arising from the exponential dimensionality of the quantum Hilbert space. However, recent PQK studies often rely heavily on extensive classical hyperparameter tuning of the radial basis function (RBF) kernel parameter and the support vector machine regularization parameter. The extensive tuning can mask the true representational power of the underlying PQKs induced by quantum feature maps, which has not been explored in detail. Thus, this work aims to expose the structural capabilities of PQKs by isolating and evaluating their native inductive bias through comparisons between untuned and tuned PQKs against the RBF baseline. Furthermore, we propose novel quantum feature maps that leverage two-qubit entanglement to capture multivariate correlations in IoT data. The results show that extensive tuning can lead to competitive performance architectures, while our proposed entanglement topologies exhibit a significantly stronger native inductive bias. This reveals that superior quantum architectural design can yield structurally better PQKs with improvements arising from quantum-mechanical properties rather than from classical hyperparameter tuning.

0.07

When to Follow the Teacher: Advantage-Based On-Policy Distillation

cs.LG cs.AIstat.ML ·2026-10-22 ·Mahdi Sabbaghi, George Pappas, Hamed Hassani, Adel Javanmard
Abstract

Online Policy Distillation (OPD) often outperforms offline methods such as Supervised Fine-Tuning (SFT) by training on trajectories sampled directly from the student. However, standard OPD trains the student to follow the teacher at every state of the trajectory, assuming that teacher guidance is always beneficial. We instead distinguish between what the teacher prefers and what actually helps the current student succeed. We introduce Advantage-Based On-Policy Distillation (AOPD), which weights teacher actions by their counterfactual advantage under the student. Teacher actions that improve the student's expected outcome are reinforced, while those that do not are downweighted. Directly computing these advantages requires trying alternative teacher actions at each step and running new rollouts from each. Instead, we show that the AOPD objective can be estimated using only standard student rollouts and their terminal rewards. Our estimator uses teacher importance weights on the student-sampled actions to provide unbiased advantage-weighted supervision, without any additional rollouts or reward evaluations. This allows AOPD to focus teacher supervision on actions that help the current student while preserving the rollout efficiency of standard OPD. Experiments on math, coding, and Medqa show gains over distillation and RL baselines in both external-teacher and self-distillation settings.

0.06

A Computationally Efficient and Rate-Optimal Algorithm for Adversarial Linear MDPs with (Aggregate) Bandit Feedback

cs.LG ·2026-10-22 ·Haolin Liu, Chen-Yu Wei, Hanbin Zhou
Abstract

We study episodic linear Markov decision processes with unknown transitions, adversarial losses, and bandit feedback. The state-of-the-art result for this setting (Liu et al., 2024) either achieves $\sqrt{K}$ regret with a computationally inefficient algorithm or achieves suboptimal $K^{3/4}$ regret, where $K$ is the number of episodes. Our work gives the first $\sqrt{K}$ regret with a polynomial-time algorithm. Our method reduces adversarial linear MDPs to adversarial linear bandits whose action set consists of policies' expected feature vectors. The central challenge is that this action set is unknown and can be estimated only through exploration. We address this challenge by constructing an approximate separation oracle for the action set that either generates executable policies from queried features, or refines an outer approximation of the action set. This approach draws inspiration from the reduction of van Erven et al. (2025) for linear contextual bandits and the oracle-efficient linear bandit algorithm of Ito et al. (2019), but additionally deals with the imprecision of the oracle. The resulting algorithm requires only aggregate loss feedback, and to our knowledge. This is the first polynomial-time algorithm for adversarial MDPs with aggregate feedback beyond the tabular case.

0.06

One Teacher, Many Views: Reducing Reference Sensitivity in On-Policy Self-Distillation

cs.LG cs.CL ·2026-10-22 ·Khouloud Saadi, Di Wang
Abstract

On-policy self-distillation (OPSD) improves reasoning by training a model on its own trajectories with dense token-level supervision from a privileged copy conditioned on a reference solution. Existing methods typically use a single teacher distribution conditioned on the complete reference. We identify a limitation of this formulation: a reference is only one realization of the reasoning process. Although it provides evidence about the answer, it also encodes an ordering, phrasing, and solution path. Full-reference conditioning can therefore produce correct targets that depend excessively on information unavailable to the student at inference. We show that replacing one valid reference with another, while holding the problem and student state fixed, can substantially alter the teacher distribution, a phenomenon we call reference-induced target instability. To address this instability, we introduce Monte Carlo Privileged Distillation (MCPD), which models privileged context as a distribution over partial reference views and distills the marginal teacher distribution. MCPD evaluates the teacher under multiple views and averages the resulting next-token distributions. We show that this probability-space average is the exact target induced by minimizing the expected teacher-to-student KL divergence over privileged views. Across mathematical and scientific reasoning benchmarks, we show that MCPD produces more reference-consistent supervision and consistently improves over full-reference OPSD.

0.06

When Forgotten Facts Return: Relearning-Aware Unlearning of Knowledge Graph Embeddings via Tangent Kernels

cs.LG ·2026-10-22 ·Parnian Fazel, Mohammad M Maheri, Hamed Haddadi
Abstract

Approximate unlearning answers a deletion request on a knowledge-graph embedding by editing entity embeddings instead of retraining from scratch. These embeddings keep being updated as new facts arrive. Under this ordinary continued training, forgotten facts recover more than in the exact-unlearning baseline, a model retrained without them and given the same continued training. We trace this relearning to entity embeddings shared by forgotten and new facts. To our knowledge, this is the first study of relearning in knowledge-graph embeddings. Relearning-robust methods developed for deep networks do not account for the graph structure underlying this recovery. We propose Kernel-Anticipated Unlearning (KAU), which uses the tangent kernel to model how continued training changes forgotten facts' scores through these shared parameters. KAU expresses this first-order response in closed form and differentiates through it inside the forgetting loss. The objective steers edits toward directions that continued training reverses less. We compare KAU with state-of-the-art knowledge-graph unlearners and relearning-robust methods under identical continued training, measuring leakage beyond the exact-unlearning baseline. Across three backbones, KAU reduces the strongest knowledge-graph unlearner's leakage by 46-63% on FB15k-237 and 50-82% on CoDEx-L, at higher retained utility.

0.05

Language, Vision & Media 231

Which Answer Did the Seventeen-Year-Old Write? A Behavioral Case Study of Author-Age Misattribution, Miscalibrated Confidence, and Non-Corrective Self-Explanation in Six Frontier Language Models

cs.CL ·2026-10-22 ·Brianne Lee
code/data
Abstract · 15 pages, 6 tables. Code and data: https://github.com/BraveAnn011/llm-author-misattribution

Three students answered the same exam prompt by hand: "What is happiness? Are you happy?" Their ages, recorded at collection, are 17, 28, and 45. The 17-year-old's answer is hedged and formally phrased. Asked to match each answer to an age and gender label, most frontier language models confidently assign it to an adult. This paper reports a structured case study of that failure across six models: Claude Fable 5, Gemini 3.1 Pro, GPT-5.6 Sol, Grok 4.5, GPT-4o (2024), and DeepSeek V4. We ran 1,081 API sessions and scored 3,225 responses across six experimental conditions that manipulate framing, disclosure format, and audience-awareness instruction. Three of six models perform at or near chance (1/3) on the matching task while reporting 73-88% mean confidence. The youngest writer's answer is misattributed to an adult in up to 90% of trials. Removing the multiple-choice scaffold collapses accuracy further: three models never place within three years of the true age once open-ended estimation is required. Institutional framing ("a community-college biology final") significantly degrades the best-performing model's accuracy; a neutral label does not. A single audience-awareness instruction significantly improves one model (+21 points) and moves another to ceiling, with no detectable effect on the three weakest models. Stated confidence is uncorrelated or inversely correlated with correctness for five of six models. After ground truth is revealed, one model names its own stereotyping in 68 of 114 explanations, yet shows the lowest accuracy and highest confidence of any model under blind presentation: articulate retrospective diagnosis that does not track the quality of its judgments. This is a three-item, single-topic, single-language case study meant to isolate a mechanism, not to estimate a population bias rate. Data, code, and raw model outputs are publicly released.

0.68

Machine-Readable Operator Manuals over the ISO 11783 Network for Offline, Multilingual LLM Assistance

cs.CL cs.AIcs.IR ·2026-10-22 ·Timo Oksanen, Julius Gun
publishedcode/data
Abstract · Accepted manuscript. 18 pages, 9 figures, 8 tables. Code and dataset: https://github.com/TUMAMX/agriquery/tree/main/ManualSync

If agricultural machinery is to support intelligent, offline assistance, then the system must access technical documentation dynamically. In this paper, the Semantic Synchronization of operator manuals over ISO 11783-based CAN buses is analyzed, defined as the process that brings the local knowledge base of a tractor-side assistant into agreement with the operator manual of the connected implement. Data transfer performance was benchmarked using Markdown, JSON, and XML; various open-weight Large Language Models (LLMs) were also evaluated in cross-lingual query tasks. The automatic evaluation of the answers was validated against four human raters. A complete manual was transferred over a physical ISO 11783 network in 31 to 89 s, depending on format and transport settings, which is comparable to the object pool upload that operators already wait for at the first coupling of an implement. Markdown produced the smallest files, while the question-answering performance was comparable across the three formats for most models. With retrieval-augmented generation, locally run models answered most questions on the English manual correctly, with no advantage of the largest model over 7B and 8B models, but performed worse on the five other language editions. Keyword retrieval failed across languages, whereas semantic retrieval provided the cross-lingual capability. The exchange requires only a new parameter group that uses the existing request and transport mechanisms of ISO 11783, which makes it a candidate for standardization. The study is a proof of concept with inference on PC-grade hardware; deployment on embedded agricultural hardware remains future work.

0.65

A Unified Framework for Image-to-3D Part Generation via Variable Granularity

cs.CV ·2026-10-22 ·Qiwen Gu, Xin Zhou, Xiwu Chen, Zhongyang Zhu, Zhuo Zhang +1 more
top venuecode/data
Abstract · Accepted at NeurIPS 2026. Code: https://github.com/Gucci233/FlexPart

Controllable 3D part generation is pivotal for digital asset creation, yet existing methods lack the flexibility to accommodate diverse user workflows requiring varying levels of control. In this paper, we introduce FlexPart, a unified image-to-3D part generation framework that supports variable-granularity geometric conditions within a single model. By seamlessly integrating heterogeneous inputs, ranging from sparse points to dense bounding boxes and masks, via gated adaptive modulation, FlexPart unlocks a powerful cross-modal synergy. We also propose an asymmetric geometric guidance mechanism that leverages strong spatial priors to align representations across granularities, improving geometric consistency and facilitating inter-task synergy. Under mask conditioning on PartVerseXL, FlexPart reduces Part CD by 16.6% relative to TripoSG + SAM3D and by 10.5% relative to OmniPart, while maintaining competitive whole-mesh quality. The code can be found at https://github.com/Gucci233/FlexPart

0.65

IQuest-Q1: Advancing Foundation Capabilities for Agentic CLI Systems

cs.CL ·2026-10-22 ·A. Jian Yang, An Huang, Bo Dan, Chen Mu, Chengran Yang +172 more
code/datacommunity pick
Abstract · IQuest Report

We present IQuest-Q1, a sparse Mixture-of-Experts (MoE) agentic foundation model with 320B total parameters (15B activated). IQuest-Q1 is developed through a three-stage process: (1) stable and efficient pre-training, combining Stable SplitNorm, Muon, and muP with a two-stage curriculum whose second stage increases the coding-data mixture, preserving a wide stability margin between safe and optimal learning rates; (2) fine-grained mid-training, expanding the data mixture across domains and formats and progressively extending the context; and (3) multi-scaffold with multi-version post-training, combining SFT, domain-specialized RL, and MOPD across various scaffolds to integrate complementary training signals and strengthen agentic coding capabilities. Furthermore, model merging combines different experts and checkpoints from different stages while mitigating conflicts among their parameter updates. The resulting IQuest-Q1 demonstrates capabilities in both coding and general tasks. We release IQuest-Q1 as an open-source model to support community research and development.

0.63

AnyGameRenderer: Entity-Consistent Generative Rendering Across Games and an Entity-Level Benchmark

cs.CV ·2026-10-22 ·Ruicong Liu, Liangyang Ouyang, Fengbo Lan, Zhixiang Wang, Kaipeng Zhang
benchmarkcode/data
Abstract

AI coding agents can now build small playable games in minutes, but the results usually look like programmer art: gray blockouts, colored geometric primitives, and flat semantic blobs. We present AnyGameRenderer, a video-to-video generative renderer that turns such footage into high-quality visuals with characters, motion detail, and effects. Throughout, it keeps the layout, camera, and every entity's position and timing intact. Our AnyGameRenderer generalizes across a wide range of genres that agents build, such as tower defense, auto-battlers, top-down survivors, turn-based battles, kart racing, and crowd runners. AnyGameRenderer takes plain video as its game-state input, with no complex representations such as engine G-buffers, per-object class labels, or low-poly proxies. Gray blockouts, colored primitives, semantic maps, and wireframes all work as input. It also preserves entity correspondence: group counts stay correct, and entities appear and disappear on the right frame. We implement AnyGameRenderer by LoRA fine-tuning MiniMax-H3 ref2va, a 33B reference-video-conditioned video generation model. Training uses ~10k paired clips from three sources: (i) multi-stream rendering data from 38 agent-made games; (ii) two-stream rendering data from Unreal Engine, with programmatically controlled crowds, count changes, and entity scales; and (iii) real gameplay recordings, with pairs reconstructed via 3D-box detection. To evaluate the entity correspondence, we propose an entity-level benchmark of 59 held-out cases with 753 entities, scored against ground-truth boxes. Compared with the base model, AnyGameRenderer better preserves entity correspondence: temporal match rises from 50.9% to 62.9%, missed count changes fall from 34.7% to 21.3%, and fewer entities are hallucinated or left unrendered. Project page: https://alaya-lab.github.io/AnyGameRenderer.

0.63

SurveyGraph: Generating Academic Surveys via Methods-Entity Knowledge Graphs

cs.CL cs.AIcs.IR ·2026-10-22 ·Yifan Yuan, Yancheng Zhu, Zuoyu Zhang, Xiaogang Peng
top venuebenchmarkcode/data
Abstract · Accepted at EMNLP 2026 Findings. 34 pages including appendices

Most automated survey generation systems retrieve papers as flat documents, limiting their ability to capture research structure. We present SurveyGraph, a framework built around a Methods-Entity Knowledge Graph (MEKG) encoding method-level follow-on, competition, and shared-resource relations. The graph supports core-paper discovery, retrieval, hybrid outlining, and evidence-gated narrative synthesis. We evaluate retrieval, semantic edges, citation grounding, collection coverage, and generation-stage components. In a controlled seven-method equal-budget benchmark, MEKG-anchored fusion balances recall and structural density, with an independent co-citation check. In a separate archived graph-native diagnostic, graph-aware retrieval nearly doubles recall over flat lexical baselines across eight runs (p = 0.004). A semantic evidence gate achieves AUC = 0.75. Audits of 160 generated sentences find 80.6% anchored-or-partial citation support. Human adjudication of 108 stratified semantic edges finds 68.5% supported-or-partial relations (binary-support Cohen's kappa = 0.64), with improves strongest (80.6%) and extends weakest (52.8%). A SurveyBench-facing collection probe reaches mean reference coverage of 0.521 across four topics with an eight-paper budget. Same-backend Qwen diagnostics on a 30-paper/74-relation graph show reduced organization signals when hybrid outlining is removed. Concurrent-system comparisons provide bounded context rather than matched-backend end-to-end evidence. The results position MEKG as an auditable structural layer for typed diagnostics, relation-specific filtering, and controlled survey generation. Code and artifacts: https://github.com/xzzvsxd/SurveyGraph.

0.59

WovenAnchor Matcher: Specialized Intra- and Inter-Image Context Modeling for Feature Matching

cs.CV ·2026-10-22 ·Zhiyang Li, Ruijiang Jin, Thibaut Klenke, Yusuke Sekikawa, Nakamasa Inoue
top venuecode/data
Abstract · Accepted at NeurIPS 2026

Semi-dense local feature matching commonly aggregates contextual information before coarse-to-fine correspondence estimation. However, intra-image and inter-image aggregation serve different roles: the former should propagate spatial evidence within each image while retaining structured two-dimensional dependencies, whereas the latter should gather information from plausible corresponding regions while limiting noise from unrelated locations. We propose WovenAnchor Matcher (WAM), a semi-dense matching framework based on role-specialized context aggregation. WAM introduces two complementary operators. WovenMamba performs intra-image aggregation through horizontal-then-vertical state-space propagation, allowing vertical updates to operate on horizontally contextualized features and thereby encouraging a structured two-dimensional receptive-field bias. AnchorCrossAttention performs inter-image aggregation by using cross-attention to gather context from plausible corresponding regions across the image pair. It performs this exchange over local representative anchors that aggregate feature evidence within each window. On MegaDepth, WAM achieves 66.1 pose AUC@5$^\circ$, improving on the JamMa baseline by 2.0 AUC points under the same evaluation setup. Ablations show that removing or replacing either specialized component reduces accuracy, providing evidence that intra-image propagation and inter-image retrieval benefit from role-specific aggregation designs. Code and pretrained weights are available at https://github.com/mmaiLab/WovenAnchorMatcher.

0.59

One-shot federated unsupervised domain adaptation with Smoothed Knowledge Distillation and teacher refinement

cs.CV cs.AI ·2026-10-22 ·Ali Abedi, Q. M. Jonathan Wu, Ning Zhang, Farhad Pourpanah
top venuecode/data
Abstract · Published in Engineering Applications of Artificial Intelligence, Vol. 167, 113946 (2026). DOI: 10.1016/j.engappai.2026.113946. Available at…

Federated unsupervised domain adaptation (FUDA) addresses the challenge of adapting models to an unlabeled target domain using decentralized source domains while preserving data privacy. However, existing FUDA methods typically require multiple communication rounds, rely on complex aggregation strategies, and often struggle with noisy pseudo labels and inconsistent source knowledge. To address these challenges, we propose SKD-ETR, a novel one-shot FUDA framework that combines Smoothed Knowledge Distillation (SKD) and Exponential moving average-based Teacher Refinement (ETR). SKD trains a student model on the target domain using smoothed soft pseudo labels generated by the ensemble of source models. This reduces overconfidence, mitigates noise, and improves robustness. ETR further refines each source model by interpolating its parameters toward the student via exponential moving average (EMA), thereby transferring target-domain knowledge back to the teachers. This bidirectional refinement enhances pseudo-label quality and student generalization without additional communication overhead. SKD-ETR eliminates complex aggregation by initializing the student model randomly and performing a single-round distillation process. Extensive experiments on OfficeHome, Office-Caltech, and DomainNet demonstrate that SKD-ETR achieves competitive performance while being communication- and computation-efficient, and robust under noisy supervision. The code for this paper is available at https://github.com/A-Abedi/SKD_ETR.

0.58

Verus-Insight: Towards Explainable Artifact Detection in AI-Generated Videos

cs.CV ·2026-10-22 ·Zhuoran Zhang, Yang Shi, Qixun Wang, Tengfei Liu, Yuqi Tang +9 more
benchmark
Abstract · 36 pages

Recent advances in video generation have made synthetic videos increasingly realistic, placing greater demands on reliable AI-generated video detection. Existing artifact-based detectors typically identify generation artifacts as evidence for source prediction, yet they may either fail to recognize subtle artifacts or reach the correct source decision without faithfully identifying the evidence underlying it. In this work, we systematically investigate how to design and train effective artifact detectors and formulate artifact detection as a structured process comprising three complementary capabilities: Inspection Planning, which determines what should be examined; Evidence Verification, which verifies whether the inspected content exhibits genuine generation artifacts; and Evidence Aggregation, which integrates the verified evidence into the final source decision. Building on this formulation, we introduce Verus-Insight, trained with human-anchored full-process supervision and progressive capability training to strengthen both fine-grained artifact detection and video-level detection. We further introduce Verus-Bench, a benchmark with fine-grained human annotations of perceptible artifacts for evaluating artifact detection. Extensive experiments across diverse benchmarks demonstrate that Verus-Insight achieves strong performance in both fine-grained artifact analysis and AI-generated video detection, with substantial gains in recovering human-perceived artifacts. Further analyses reveal that the three capabilities are complementary yet distinct, and that Verus-Insight more reliably converts relevant inspection targets into supported artifact evidence without increasing spurious artifact claims.

0.56

Vision Transformers for Estimating Irradiance Using Data Scarce Sky Images

eess.IV cs.AIcs.CVcs.LG ·2026-10-22 ·David Hamlyn, Sunny Chaudhary, Tasmiat Rahman
publishedhighly cited
Abstract

Accurate estimation of diffuse horizontal irradiance (DHI) is critical for optimising photovoltaic system performance and energy forecasting yet remains challenging in regions lacking comprehensive ground-based instrumentation. Recent advancements using Vision Transformers (ViTs) trained on extensive sky image datasets have shown promise in replacing costly irradiance measurement equipment, but the scarcity of long-term, high-quality sky imagery significantly restricts practical implementation. Addressing this critical gap, this study proposes a novel dual-framework approach designed for data-scarce scenarios. First, calculated atmospheric parameters, including extraterrestrial irradiance and cyclic time encodings, are integrated to represent sky conditions without utilising any instrumentation. Next, a sequential pipeline initially predicts synthetic global horizontal irradiance (GHI) and uses it as a feature, to refine DHI estimation. Finally, a dual-parallel architecture simultaneously processes raw and overlay-enhanced fisheye sky images. Overlays are generated through unsupervised, physics-informed cloud segmentation to highlight dynamic sky features. Empirical validation is performed using data from the Chilbolton Observatory, chosen for its temperate climate and frequent cloud variability. To simulate data-scarce conditions, models are trained on a single month (e.g., January) and evaluated across a temporally disjoint, full-year test set. Under this setup, the sequential and dual-parallel frameworks achieve RMSE values within 2-3 W/m^2 and 1-6 W/m^2, respectively, of a state-of-the-art ViT trained on the complete dataset. By combining physics-informed modelling with unsupervised segmentation, the proposed method provides a scalable and cost-effective solution for DHI estimation, advancing solar resource assessment in data-constrained environments.

0.56

CellTrack-Bench: Microscopic Cell Tracking Benchmark for Vision-Language Models

cs.CV ·2026-10-22 ·Ilan Theodoro, Jordão Bragantini, Loïc A. Royer
benchmarkcode/data
Abstract

Large language models have shown outstanding performance on text-first tasks such as coding, math, and writing. Recent efforts are shifting models' training to include scientific problems that often live in highly specialized domains beyond text, such as protein structure and biological images. However, in these scenarios, their general-purpose accuracy compared with specialized algorithms remains in question. One such domain is cell tracking, which requires spatiotemporal awareness of cell dynamics. To evaluate this, we created a new benchmark, CellTrack-Bench, which assesses Vision-Language Models' (VLMs) cell-tracking-related performance against specialized trackers. The full benchmark draws from 15 standard cell-tracking datasets across four imaging modalities and comprises 1,537 problems corresponding to 9,041 cell-linking decisions. We selected a hard subset requiring greater reasoning effort by identifying failures shared across distinct algorithmic trackers. With this proposed dataset, we: (i) evaluated different environment harnesses for VLMs; (ii) benchmarked 9 frontier closed-source models and 13 open-weight models; and (iii) provided a detailed analysis of their failure modes. Code, the stored model outputs, and a pipeline that regenerates the benchmark from the source datasets are available at https://github.com/royerlab/celltrack-bench.

0.54

WikiOriginQA: Automated Construction of a QA Dataset with Knowledge Origins for Analyzing Cultural Bias

cs.CL ·2026-10-22 ·Kento Haneda, Yosuke Kishinami, Ryo Fujii, Keisuke Sakaguchi, Makoto Morishita
top venuedatasethot direction
Abstract · Accepted to AACL-IJCNLP 2026 (9 pages, 2 figures, 10 tables)

Existing datasets on cultural knowledge have been typically created by human annotators familiar with each culture. However, the process requires considerable human effort, presenting a significant challenge to scalability. To address this challenge, we introduce WikiOriginQA, a QA dataset automatically constructed by objectively representing the origins of knowledge through the creation dates of different Wikipedia language editions. Experimental results with the created dataset confirm that LLM performance tends to be biased toward Anglophone cultures. Furthermore, the results suggest the importance of knowledge propagation into the Anglosphere. These findings demonstrate the validity of our automated approach, enabling cultural bias evaluations with superior scalability across diverse languages without extensive manual annotation.

0.53

GigaBench: Benchmarking Learnable Representations for Gigapixel Image Compression and Super-Resolution

cs.CV ·2026-10-22 ·Matteo Fusconi, Alex Costanzino, Pierluigi Zama Ramirez, Luigi Di Stefano
publishedbenchmarkcode/datahot direction
Abstract · Project page at https://matteofusconi.github.io/GigaBench Accepted at WACV2027

Learnable Representations (LRs) can fit parametric families of functions to diverse kinds of signals. These emergent neural-native data formats are gathering interest in the scientific community due to their compactness, resolution-agnosticism and amenability to direct neural processing. On the other hand, gigapixel images are becoming increasingly widespread across a variety of scientific, cultural, and industrial domains. In this paper, we present GigaBench, the first benchmark and experimental evaluation focused on deploying LRs with gigapixel imagery. GigaBench includes images collected, acquired, and synthetically rendered across diverse domains, and addresses image compression and super-resolution. We adapt a representative set of LRs, spanning Implicit, Hybrid, and Explicit methods, so as to operate reliably at the gigapixel scale on consumer hardware. Through systematic evaluation across reconstruction quality, model size, compression and inference times, we identify consistent trends across methods and suggest actionable guidance for future investigation.

0.52

GLOW-FM: Efficient Generative Low-Light Image Enhancement with One-Step Latent Flow Matching

cs.CV ·2026-10-22 ·Ariel Lapid, Yuval Haitman, Reuven Peretz, Arnon Netzer, Idit Diamant
publishedcode/data
Abstract · WACV 2027

Low-light image enhancement (LLIE) aims to recover well-exposed and perceptually faithful images from severely degraded observations. Recent diffusion-based generative methods achieve strong perceptual quality by leveraging powerful image priors, but their iterative sampling incurs substantial computational cost, limiting practical deployment. Despite recent efficiency gains in generative LLIE, high-quality enhancement with a single function evaluation and low overall computation remains challenging, particularly under diverse real-world low-light conditions. We introduce GLOW-FM, a one-step latent flow-matching framework for efficient generative LLIE. GLOW-FM performs conditional transport in the latent space of a frozen Tiny-VAE and leverages Terminal Velocity Matching (TVM) to enable enhancement with a single function evaluation. Matched comparisons with alternative one-step objectives support TVM as an effective design choice, and, to the best of our knowledge, this is its first adaptation to image restoration and specifically to LLIE. Extensive experiments on challenging real-world benchmarks demonstrate that GLOW-FM achieves competitive image quality and state-of-the-art cross-dataset generalization while requiring only a single function evaluation and 34 GFLOPs for the small variant. These findings highlight the potential of GLOW-FM to bring generative LLIE closer to practical deployment through single-step inference and compact latent modeling. Code will be made publicly available upon publication.

0.51

RetiJev: Fast and Source-Grounded Ophthalmic Decision Models

cs.CL cs.AI ·2026-10-22 ·Bo Liu, Lianyu Wang, Meng Wang, Ching-Yu Cheng, Huazhu Fu
benchmarkcode/data
Abstract

Ophthalmic clinical decision support requires fast, accurate, and privacy-preserving inference, yet autoregressive large language models (LLMs) incur substantial latency and computational overhead for decisions among predefined actions. We therefore investigate a decision-oriented alternative to generative reasoning. Specifically, we first introduce RetiJev-Data, a source-grounded ophthalmology corpus comprising 78,158 clinical-decision and diagnosis questions derived from articles in PubMed Central and a MeSH vocabulary covering 531 eye diseases, and 50,074 guideline-grading questions spanning 11 grading systems, along with a 24,662-question evaluation benchmark. Based on this, we develop RetiJev, a family of 0.8B-9B decision models that directly score candidate answers without autoregressive decoding. Across all model sizes, RetiJev outperforms both a 27B LLM and the commercial Jev 1.13 decision model on guideline grading by 10.4-23.0 points. RetiJev-9B further achieves the highest accuracy on clinical decisions, improving over both baselines by 5.4-6.5 points while matching them on diagnosis. It answers a case in 158 ms on one GPU, 3.5x faster than LLM generation and 470x faster than reasoning. Models are released at https://github.com/Awenbocc/RetiJev.

0.50

Formal-Informal Mathematical Alignment through and beyond Human Annotations

cs.IR ·2026-10-22 ·Noah Gießing, Ankit Satpute, Hamid Rahkooy, André Greiner-Petter, Moritz Schubotz
top venuedatasetcode/data
Abstract · Accepted at: Proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM '26), November 7-11, 2026, R…

Formalized mathematics libraries such as Lean's \texttt{mathlib} provide machine-verifiable representations of mathematical knowledge, including definitions, theorems, and their relationships. If effectively connected to informal mathematical resources, formal mathematics could help uncover previously implicit links between mathematical concepts and improve access to existing formalizations. However, connections between informal and formal mathematics libraries are currently sparse and unsystematic. In this paper, we make first strides towards better formal integration. We construct a dataset linking informal statements from the Stacks Project to their formalizations in \texttt{mathlib} by exploiting expert annotations. We evaluate retrieval, LLM-based reranking, and LLM-assisted verification to discover new unannotated informal-to-formal links. To facilitate exploration and reuse of formalized knowledge, we integrated the newly discovered alignments into the MaRDI knowledge graph, enabling mathematicians to find formal counterparts to informal mathematical content. We discuss the difficulties and challenges in using LLMs for linking formal and informal mathematical content. All code and data necessary for reproducing this paper can be found at: https://github.com/AnkitSatpute/Formal-Informal-Mathematical-Alignment

0.49

GSHeadRelight: Fast Relightability for 3D Gaussian Head Synthesis

cs.CV ·2026-10-22 ·Henglei Lv, Bailin Deng, Jianzhu Guo, Xiaoqiang Liu, Pengfei Wan +2 more
publishedcode/datahighly cited
Abstract

Relighting and novel view synthesis of human portraits are essential in applications such as portrait photography, virtual reality (VR), and augmented reality (AR). Despite recent progress, 3D-aware portrait relighting remains challenging due to the demands for photorealistic rendering, real-time performance, and generalization to unseen subjects. Existing works either rely on supervision from limited and expensive light stage captured data or produce suboptimal results. Moreover, many works are based on generative NeRFs, which suffer from poor 3D consistency and low real-time performance. We resort to recent progress on generative 3D Gaussians and design a lighting model based on a unified neural radiance transfer representation, which responds linearly to incident light. Using only in-the-wild images, our method achieves state-of-the-art relighting results and a significantly faster rendering speed ($\times12$) compared to previous 3D-aware portrait relighting research. Project page: http://www.geometrylearning.com/GSHeadRelight/.

0.49

HiSTEVE 1.0: Harnessing Human-Interface VLA Agents for Open-World Games

cs.CV ·2026-10-22 ·Zecong Tang, Zian Guan, Yuncheng Guo, Zhanqiu Zhang, Yiwen Guo +1 more
community pick
Abstract

Building virtual embodied agents that operate in open-ended games like human players is a long-standing goal requiring both long-horizon autonomy and reliable native-interface control. As a procedurally generated open world, Minecraft provides a natural testbed for studying the integration of these capabilities. Existing agents typically provide either native visuomotor control without long-horizon autonomy or high-level planning executed through environment APIs or predefined skills. To bridge this gap, we propose HiSTEVE, a hierarchical VLA agent with a capability-aware agent harness. At the orchestration level, we equip the harness with structured memory for tracking progress and a VLM planner for selecting subtasks within the VLA's capabilities. To reliably determine subtask completion, we further introduce a policy-coupled task completion state detector that conditions on VLA policy context to signal candidate completion. At the control level, a single VLA executes all subtasks with vision-only perception and native keyboard-and-mouse actions. To mitigate interference between heterogeneous GUI and non-GUI interactions, we introduce regime-specific RMSNorm branches and separate mouse-movement token embeddings. We also train the VLA through world knowledge injection, imitation learning, and reinforcement learning with environment feedback to improve reliability and recovery. HiSTEVE achieves state-of-the-art atomic-task performance, scoring 85.0% on MineStudio Bench and a 37.4% average success rate on OpenHA, while also completing long-horizon tasks such as crafting an iron pickaxe from scratch like human players.

0.49

Evaluation of Uncertainty-Aware Multi-Software Ensembles for Hippocampal Segmentation

cs.CV ·2026-10-22 ·Gabriel Oliveira-Stahl, Anna Schroder, James Moggridge, Hamza A. Salhab, Caroline Micallef +4 more
top venue
Abstract · Accepted for publication at the Journal of Machine Learning for Biomedical Imaging (MELBA) https://melba-journal.org/2026:030

Accurate hippocampal segmentation can be a useful tool for diagnosing and monitoring neurological conditions such as Alzheimer's disease and epilepsy. While numerous automated segmentation methods exist, their clinical adoption remains limited. Reliable uncertainty assessment can enhance trust and facilitate clinical translation. This study evaluates ensembles of five heterogeneous hippocampal segmentation methods - InnerEye, ASHS, FastSurfer, HippoSeg, and FreeSurfer - across two dementia datasets and one epilepsy dataset. The sub-ensemble containing InnerEye, FastSurfer, and HippoSeg emerged as both accurate and efficient, highlighting the feasibility of balancing computational cost and performance. Additionally, ensemble-derived uncertainty quantification with sample variance, mutual information, and predictive entropy is shown to reduce inaccurate segmentations by flagging low-confidence cases, potentially providing a mechanism for automatically escalating ambiguous cases for expert assessment. Lastly, this study explores the downstream clinical utility of ensemble-derived uncertainty, finding a significant correlation with motion artefact severity and a modest contribution to Alzheimer's disease classification.

0.49

When LLMs Forget What They Believe: Characterizing State Integrity Collapse in Long-Horizon Dialogues

cs.CL cs.AI ·2026-10-22 ·Yancheng Zhu, Yifan Yuan
top venuebenchmarkcode/data
Abstract · Accepted at EMNLP 2026 Main Conference. 17 pages including appendices

Large language models can repeat an instruction while failing to keep obeying it. We study this recall-compliance dissociation in long-horizon dialogue and define State Integrity Collapse: a model preserves a stated constraint in memory but stops applying it to later turns. We introduce StateTrack-Bench, a controlled suite of conversations covering numerical, format, role/persona, factual-memory, and logical-rule constraints. Across seven provider-pinned model snapshots and five dialogue-depth probes, strict joint retention ranges from 62.4% to 100.0%, with degradation concentrated in symbolic constraints and in one especially brittle model. A forked recall audit shows that 88.6% of observed failures are behavioral-drift failures rather than ordinary forgetting; excluding the brittle outlier preserves this diagnosis (84.4% of failures). Extended evaluations at depths up to 50 turns confirm that the dissociation persists beyond the core benchmark, while Structured State Anchoring recovers up to 24.4 points on immutable constraints. The results show that reliability cannot be inferred from single-turn instruction following alone; evaluation must track whether remembered constraints remain operational as dialogue state grows. Code and artifacts are available at: https://github.com/xzzvsxd/StateTrack.

0.49

F2G-Pose: Geometry-Aware Foundation Feature Lifting for Direct RGB-D Category-Level Object Pose Estimation

cs.CV ·2026-10-22 ·Jinyu Zhang, Haitao Lin, Jiashu Hou, Xiangyang Xue, Yanwei Fu
top venuecode/data
Abstract · Accepted at NeurIPS 2026. 22 pages, including the complete technical appendix

This paper studies category-level RGB-D object pose estimation, which recovers an object's 3D rotation and translation without instance-specific CAD models or reference views. This task is highly ambiguous due to intra-class variations, symmetries, and partial observations. Current solutions struggle to balance reliability and efficiency: methods based on intermediate structures are prone to error propagation, while diffusion models suffer from high inference latency and complex post-hoc candidate extraction. To address this, we present F2G-Pose, a real-time direct, single-pass framework that predicts category-level pose, metric size, and camera-frame completed shape from a single RGB-D observation in one feed-forward pass. F2G-Pose lifts dense visual foundation features to partial point clouds, encodes local geometry with DGCNN tokenization, and integrates object-level structure with a Mixture-of-Experts Transformer. Joint size prediction and camera-frame shape completion encourage metric consistency through dense geometric supervision. With its task-specific network trained only on pose-annotated synthetic SOPE data and its frozen RADIO backbone externally pretrained, F2G-Pose achieves state-of-the-art accuracy on SOPE and ROPE while running at 28 FPS for pose-only inference, and further demonstrates strong camera-frame geometry on HouseCat6D and real-object transfer on HANDAL. Analyses of latency, robustness, shape completion, and GenPose++ candidate behavior further demonstrate the efficiency and reliability of F2G-Pose, with real-world manipulation experiments demonstrating deployment potential under segmented RGB-D inputs. The project page, code, and pretrained models are made available at https://jrryzh.github.io/F2G-Pose/.

0.48

Data-Driven Vision-Language Models for Remote Sensing: A Survey

cs.CV cs.LG ·2026-10-22 ·Yue Zhou, Shujun Zhao, Xue Yang, Ruigang Li, Tianwen Zhang +5 more
publishedsurveyhighly cited
Abstract · 44 pages, 15 figures

During the deep learning era, innovations in remote sensing (RS) vision models primarily focused on optimizing network architectures for specific tasks and conducting end-to-end training. However, with the advent of the Vision-Language Models (VLMs) era, constrained by substantial computational resource demands, the mainstream research paradigm has shifted towards adapting general-purpose VLMs using RS vertical domain data for fine-tuning, rather than modifying their core architectures. This transition marks a change in the innovation pathway from model-centric to data-centric. Despite the increasing importance of data, the academic community still lacks a systematic, data-centric survey to guide researchers in constructing high-quality RS vision-language datasets for various application scenarios and in conducting effective model fine-tuning and performance evaluation based on these datasets. To address this gap, this paper presents a systematic review of the development status of RS datasets tailored for VLMs. It further provides a comprehensive practical guide for RS researchers in the large model era, covering the entire pipeline from data construction to fine-tuning and evaluation. This paper aims to encourage the academic community to build more diverse and higher-quality vision-language datasets for specialized RS sub-fields. Ultimately, the goal is to converge these efforts into a comprehensive RS vision-language corpus that offers extensive task coverage and a wide geographical distribution, thereby laying a solid data foundation for advancing towards general artificial intelligence in RS.

0.48

Joint or Per-Option? Compact Encoders for Closed-Set Decisions from 17 Million to 1 Billion Parameters

cs.CL cs.LG ·2026-10-22 ·Tardelli Ronan Coelho Stekel
benchmarkcode/data
Abstract · 18 pages, 8 figures, 8 tables. Code: https://github.com/tardellirs/crivo-decision-encoders ; models: https://huggingface.co/tardellirs

Many language-model pipelines end in a closed-set decision, such as selecting a tool. Prompting a large language model for it is slow and order-sensitive. We ask how small an encoder for such decisions can be and whether it should read the options jointly or separately. In a three-seed study at 32 million parameters, per-option scoring beats joint scoring on a held-out sample of a 37-benchmark decision suite by 5.8 points of chance-corrected skill at matched optimizer steps and by 2.8 at approximately matched training compute; two external benchmarks agree, though not significantly on their unseen tasks. Crossing reader with initialisation shows that the advantage requires a reranker initialisation, whose query-candidate pair input matches per-option reading; from the raw encoder the two readers are indistinguishable. Per-option scoring needs 5 to 13 times more encoder passes: at most twice the median latency, but more than four times the mean. Of six released joint encoders (17 million to 1 billion parameters), the 68- and 150-million-parameter models exceed a 435-million-parameter DeBERTa-v3 zero-shot classifier by 4.9 and 5.9 accuracy points on 209 overlap-audited System1-Arena tasks and are at least on par with it on the 145 tasks unseen in training. The smallest answers a median request in 14 milliseconds on 16 processor threads, against 728 for the classifier. Reordering options changes 12 to 23% of the joint models' answers (per-option scoring: none). Up to 68 million parameters, they lead their size range on a public leaderboard's private, mainly in-distribution tests.

0.48

When More Metadata Misleads: Probing Field-Level Grounding in LLM-Generated Museum Descriptions

cs.CL cs.AIcs.DLcs.LG ·2026-10-22 ·Clemencia Siro, Laura Hollink
top venue
Abstract · Accepted to EMNLP 2026 Findings

Metadata-conditioned generation is often framed as a data-to-text task in which additional input fields should improve grounding. Museum description generation challenges this assumption because catalogue records are heterogeneous, partially visual, and unevenly documented across geographic regions. We study how metadata composition shapes the generation of artefact descriptions through a two-phase framework. Phase 1 analyses natural metadata variation in a large museum collection, where field count, field type, image availability, and Place co-vary. Phase 2 introduces controlled field probes that hold object identity fixed while varying metadata quantity, field type, and image access. Our results show that metadata quantity alone does not explain generation behaviour. Field type determines whether models ground, omit, or infer information, while images can both support visual grounding and increase unsupported interpretation. More broadly, our findings show that metadata-conditioned data-to-text evaluation should assess field- and claim-level support, rather than relying only on aggregate input coverage or reference similarity.

0.46

UniGuard: Multi-View Guardrail Learning for Cross-Interface Safety Generalization

cs.CL ·2026-10-22 ·Viet-Thanh Pham, Thuy-Trang Vu, Junae Kim, Rollin Omari, Dinh Phung +2 more
Abstract · 31 pages

Guardrail models are increasingly used as safety layers for LLMs and agentic systems, but most are trained on fixed prompt or response interfaces and become unreliable when the same safety case appears as a conversation, agent trajectory, multi-agent trace, or policy-conditioned input. We propose UniGuard, a training framework that learns interface-invariant safety representations from multiple semantically aligned views of the same safety case. We also construct the UniGuard-FT dataset by transforming 1,000 public safety cases into prompt, response, multi-turn conversation, agent-trace, and multi-agent-trace views, and train UniGuard-0.6B with classification, supervised contrastive learning, and asymmetric confidence calibration losses. Across unseen-interface benchmarks ATBench, PolicyGuardBench, and DynaBench, UniGuard-0.6B improves over its pretrained guardrail by 45.12 F1 points on average and reduces expected calibration error by 35.41 points, while remaining competitive with much larger guardrails such as Claude-Sonnet-4.5 and GPT-5.2 - despite being a 0.6B model and trained only with synthetic data.

0.46

PerMix: Anomalous sound detection by predicting the perturbation ratio of mixed samples

eess.AS ·2026-10-22 ·Jihoon Choi, Jung-Woo Choi
benchmarkcode/data
Abstract · 5 pages, 1 figure, submitted to ICASSP 2027

In anomalous sound detection based on self-supervised learning, the performance of the detection model largely depends on the selection of the pretext task through which the characteristics of normal data are learned. A common pretext task is to mix samples from different classes and classify the mixed sample as its target class with an angular margin loss. However, this supervision becomes mismatched when the target sample is the minority component of the mixture and the margin remains fixed irrespective of the mixing ratio. To address these issues, we propose a pretext task that directly predicts the perturbation ratio of mixed samples by reformulating the logit so that the angular margin is scaled by the perturbation ratio. In addition, we adopt a supervised contrastive loss modified for mixed samples, in which positive pairs are formed according to the dominant class. Experiments show that our approach achieves state-of-the-art results on the DCASE 2020 Task 2 benchmark. Our code is available at https://github.com/kaistss/PerMix.

0.44

Mixing Matters: Evidence Position Bias Across Sequence Mixers in Long-Context Question Answering

cs.CL cs.AIcs.LG ·2026-10-22 ·Aman Behera, Namit Solanki, Mehul Anand
publishedcode/data
Abstract · Accepted at the 2nd Workshop on Principles of Generative Modeling (PriGM), NeurIPS 2026. 9 pages. Code: https://github.com/beingamanforever/…

Long-context models often use answer-bearing evidence less effectively when it appears away from a prompt boundary. We compare this position bias across Transformer, state-space, and hybrid Mamba architectures by moving one gold document through a fixed set of nine distractors. The evaluated Transformers show a beginning advantage that the evaluated state-space models lack, while both families favor evidence near the end. A matched pure-versus-hybrid Mamba comparison points in the same direction but remains statistically uncertain. The family difference appears among models that can answer the task. Transformer primacy reappears on synthetic retrieval, where state-space ceiling prevents a family comparison. A corpus swap within one state-space architecture changes overall accuracy without a detectable change in curve shape. Prompt changes can create or nearly remove a beginning advantage. Position remains decodable across architectures even when response curves differ, and attention concentration near initial tokens covaries with beginning sensitivity in the tested Transformer family. These exploratory results show that effective context use depends on the evaluated architecture family, capability, task, and prompt, without isolating the sequence mixer as the sole cause.

0.44

IndoNutriQA: An Indonesian Clinical Nutrition Question Answering Benchmark

cs.CL ·2026-10-22 ·Tirana Noor Fatyanosa, Ummu Ditya Erliana, Mahendra Data, Muhammad Nurwegiono, Eleazar Tadeo Eman +1 more
benchmarkcode/data
Abstract · 26 pages (9 pages main text, plus limitations, ethics, references and appendix), 5 figures, 18 tables. Under review at EACL. Data and code: …

Clinical Question Answering (QA) underpins Clinical Intelligent Decision Support Systems, but progress is severely limited for low-resource languages such as Indonesian by the absence of public domain-specific data. We introduce IndoNutriQA, an expert-curated Indonesian clinical QA benchmark of 911 question-answer pairs built from real clinical-nutrition reports. Rather than a large-scale training corpus, it is positioned as a high-quality evaluation resource: every pair is grounded in a clinically relevant context, enriched with structured metadata (anthropometric assessment, medical diagnosis, document provenance, chunk segmentation). Eight standardised question types following the Nutrition Care Process (NCP), including the Problem, Etiology, Sign and Symptom (PES) diagnosis statement, are answered for every report, which was written by a graduate-level dietetic intern, by one of three nutritionists. Answers are free-form clinical prose rather than extractive spans, and an expert validation study finds 6-10% of reference answers contestable. The dataset supports generative, extractive and retrieval-augmented QA as well as zero- and few-shot LLM evaluation. We report fine-tuned baselines (IndoBERT variants, mT5, RAG) with 5-fold cross-validation, learning curves and bootstrap 95% confidence intervals, and benchmark eight contemporary LLMs zero- and five-shot. We further calibrate our LLM-as-judge rubric against three expert raters (one RD, two nutritionists), finding it human-comparable on factuality and completeness but unreliable on clinical safety, and a human error analysis shows it misses unsupported content that the raters identify. IndoNutriQA is released to foster research in Indonesian clinical NLP and nutritional decision support.

0.44

DeXplicit: Making Physical State Explicit in Dexterous Manipulation World Models

cs.CV ·2026-10-22 ·Adam Lee, Mutian Xu, Yukang Cao, Shulin Tian, Ziwei Liu
code/data
Abstract · Project page: https://dexplicit.github.io

Before acting on an object, a robot needs to anticipate how its hand motion will change the physical world and what the resulting scene will look like. Existing approaches typically entangle action-conditioned dynamics, camera motion, and visual synthesis within a video generation model, or represent dynamics as point trajectories without explicit object structure. We introduce DeXplicit, a dexterous manipulation world model that instead predicts how the world evolves explicitly in metric 3D before rendering its visual consequences. Given an initial scene and a future hand trajectory, a lightweight 3D dynamics network predicts object-centric motion and deformation, providing structured and geometrically grounded future world states. These predictions then serve as a strong scaffold for video generation: appearance is transported through the predicted 3D motion and projected into future views, allowing the renderer to focus on completing uncertain and newly visible content rather than implicitly learning world dynamics from pixels. This explicit-3D-first decomposition also enables efficient causal generation, requiring only two denoising steps per block while conditioning on generated history. Across HOT3D, OakInk2, and PhysTwin, DeXplicit reduces moved-point error by 27-40% relative to PointWorld and improves object fidelity IoU+ by 47% on HOT3D under a shared renderer. Our design with two-step causal generation achieves a 57x speedup from our 14B bi-directional configuration in video synthesis.

0.43

What the Query Remembers: Query Abstraction Governs Lexical and Dense Retrieval of Classical Arabic Poetry

cs.CL cs.AIcs.IR ·2026-10-22 ·Omar Farouk Zouak, Selma Mani, Larbi Said Chikh, Raid Abderrezak Ouahioune
benchmarkcode/data
Abstract

How well a poetry archive can be searched depends on what the user still remembers, not only on what the model understands. We show this for Classical Arabic poetry, benchmarking nine dense encoders and BM25 over two corpora, three indexing modes and two evaluation protocols. Which retriever wins is governed by how much of a poem's wording the query still carries. On tip-of-the-tongue queries BM25 leads every dense encoder, including an 8B state-of-the-art model (0.746 vs. 0.717 Recall@5); on abstract queries it falls to eighth out of ten (0.392 vs. 0.748). Editing the wording of the same queries shows that this ranking largely reflects query composition: adding one remembered half-verse raises BM25 by 38 points, rewriting the quoted words by hand leaves BM25 level with the 8B encoder, and BM25's sensitivity to remembered wording is significantly steeper than that of the strongest dense encoder. The same axis controls explanation augmentation, whose effect swings from +37 to -39 points. The mechanism splits by system type: lexical containment predicts BM25's behaviour, while embedding proximity predicts dense encoders. Finally, the field's two standard protocols, known-item retrieval and thematic relevance, dissociate on the same systems, a result reproduced by two blind human annotators and by two open-weights judges, one of them under a rubric with no literal-match criterion. A benchmark that reports one protocol, or one query register, can give a materially wrong picture. Code and data are available at https://github.com/Ellzo/what-the-query-remembers.

0.43

StateBridge: Phase- and Track-Grounded Reasoning for Synthetic-to-Real Traffic Video Captioning and VQA

cs.CV ·2026-10-22 ·Blessing Agyei Kyem, Richard Dzinyela, Joshua Kofi Asamoah, Eugene Denteh, Andrews Danyo +3 more
published
Abstract · 25 pages, 5 figures, ECCV conference

Traffic safety videos provide rich evidence for understanding interactions between pedestrians and vehicles, but synthetic-to-real evaluation makes this setting challenging because models must learn from simulated scenes and generalize to real-world traffic. In this work, we present StateBridge, a traffic video-language framework designed to operate under this synthetic-only training protocol. Our approach uses phase-aware clip construction, pedestrian and vehicle track summaries, and structured prompts to ground vision-language models in stable traffic evidence rather than appearance alone. We fine-tune public vision-language models on synthetic captioning and visual question answering examples, then combine their outputs through fact-locked caption fusion and evidence-diverse VQA consensus to improve consistency across actor roles, behavioral phases, and question types. StateBridge achieves an official S2 score of 55.4679 on AI City Challenge 2026 Track 2 using only synthetic training data. These results demonstrate the effectiveness of structured evidence transfer for traffic safety captioning and visual question answering.

0.43

Sensorimotor Alignment: A Platonic Proxy for Embodied Policy Success

cs.CV cs.LGcs.RO ·2026-10-22 ·Yunfei Xie, Howard H. Qian, Shiyi Lan, Kaiyu Hang, Chen Wei
code/data
Abstract · 28 pages. Project page and code available at https://yunfeixie233.github.io/s2/

Robot foundation models are typically evaluated through costly closed-loop rollouts, motivating low-cost offline proxies for model capability. The Platonic Representation Hypothesis (PRH) suggests that as representations improve, modalities such as vision and language converge toward a shared statistical model of reality. We ask whether motion trajectories form another such modality: as embodied models become more capable, should their sensory representations converge toward the structure of physical motion? To test this, we propose Sensorimotor Alignment ($S^2$), a training-free offline metric that measures how well a model's sensory representation aligns with the geometry of physical motion. We find that $S^2$ correlates positively with downstream robot task success across model families, making it a low-cost offline proxy for prioritizing models before closed-loop evaluation. $S^2$ can also identify promising vision-language model (VLM) backbones before costly fine-tuning into vision-language-action models (VLAs). Beyond this correlation, $S^2$ offers representation-level explanations for phenomena observed in VLAs and world action models (WAMs). With $S^2$, we extend PRH to robotics and make sensorimotor alignment measurable, offering a low-cost, representation-level view of robot foundation models that closed-loop success rates alone cannot reveal.

0.43

StackGen3D: Stable 3D Block Stack Generation from Multiview Silhouettes

cs.CV cs.RO ·2026-10-22 ·Xiaopeng Zhang, Luzhe Sun, Yiqing Xu, Matthew R. Walter
published
Abstract · A poster of this work is publicly available on the website of the Creativity in Assembly Workshop at IROS 2026 (non-archival)

Generating physically stable 3D structures from visual specifications requires resolving geometric ambiguity while accounting for physical interactions among objects. We present StackGen3D, a framework for generating stable block assemblies conditioned on a given block inventory and three orthographic silhouettes. Our approach consists of two stages. First, we develop a physics-informed graph neural network (PI-GNN) that models support relations between blocks and serves as a fast stability predictor for curating a diverse dataset of stable structures. Second, we train a conditional flow-matching model on this dataset to capture the multimodal distribution of block configurations consistent with the target silhouettes. At inference time, lightweight geometric selection and refinement reduce inter-block penetration without physics simulation. Preliminary experiments show that the PI-GNN generalizes from simple stacks to complex structures substantially better than a parameter-matched multilayer perceptron. For structure generation, Best-of-10 selection improves physical stability from 45.3% to 72.4% while achieving a silhouette IoU of 0.970. These results highlight the potential of combining relational physical reasoning, scalable data curation, and conditional generative modeling for creative 3D assembly. Project website: [https://ripl.github.io/StackGen3D-web/](https://ripl.github.io/StackGen3D-web/)

0.42

DRILL: Training World Models to Improve Policies, Not Predict Pixels

cs.CV ·2026-10-22 ·Zhaolu Kang, Tailong Luo, Lei Wei, Guangyuan Dong, Sihan Liu +7 more
published
Abstract · NeurIPS 2026

World models are increasingly used to reduce environment interaction when post-training vision-language-action (VLA) policies with reinforcement learning. Yet they are typically trained to predict pixels one step ahead, even though their role here is not merely to render visually faithful videos but to produce learning signals from imagined rollouts that improve the policy. We show that this mismatch is substantial: lower pixel MSE can produce worse policy gradients, and a world model trained for policy utility can yield better downstream policies despite 17% higher pixel MSE on the next frame. We propose the Downstream-Return Imagination Learning Loop (DRILL), a bilevel framework that updates the world model to maximize the policy's return after an inner GRPO update on imagined rollouts. DRILL has two instantiations. DRILL-VWI is a closed-form surrogate, requiring no additional simulator rollouts, that reweights prediction errors by policy visitation, the group-relative GRPO advantage, and model confidence. DRILL-IMG is a full meta-gradient method that differentiates through the inner update using truncated Hessian-vector products. We show that DRILL-VWI recovers the local one-step component of the DRILL-IMG meta-gradient under score-compatibility and small-residual assumptions. Across five manipulation simulators and two VLA backbones, DRILL-IMG raises mean success rate by 13.1 points over WMPO and 7.9 points over RLVR-World, and complements RLVR pretraining rather than competing with it. More broadly, we believe video world models for VLA reinforcement learning can be productively trained by the policy gradients they produce, not only by pixel-level prediction.

0.42

FedBST: Bilateral Self-Training for Unsupervised Federated Classification

cs.CV ·2026-10-22 ·Aijing Yu, Kuangpu Guo, Jian Liang, Xiao-Yu Zhang
Abstract · ICME 2026

Unsupervised federated learning with Vision-Language Models (e.g., CLIP) offers a promising solution for label-free classification, yet its performance often degrades under severe non-IID shifts. A pioneering method, FedCoPL, typically relies on self-training strategies that exclusively prioritize high-confidence predictions for pseudo-labeling. However, this selective approach neglects the information of ambiguous samples (often referred to as ``waist" samples), leading to the underutilization of data and biased decision boundaries. To address this limitation, we propose FedBST, a bilateral self-training framework that effectively exploits both high-confidence and ambiguous data. Specifically, for local optimization, we introduce uncertainty-aware confidence penalization (UACP). UACP targets ambiguous samples by penalizing overconfident predictions to implicitly promote higher uncertainty, thereby preventing erroneous pseudo-labeling. At inference, we employ a visual-text ensemble strategy. This approach aggregates uncertainty-aware local visual prototypes and ensembles them with textual prototypes, achieving lightweight adaptation to local distributions via prototype-based calibration. Experiments demonstrate that FedBST achieves consistent improvements over existing methods with standard federated aggregation protocols.

0.42

Kolibri: A Sovereign European Model on the Pareto Frontier

cs.CL cs.AIcs.LG ·2026-10-22 ·Aleph Alpha, :, Moritz Althaus, Artur Baranowski, Michael Barlow +63 more
Abstract

This report introduces Kolibri, an English-German mixture-of-experts transformer with 78.1B total parameters and 3.46B active parameters per token, released as open weights under the Apache 2.0 licence. We built Kolibri for sovereign and specialised deployment, with a particular focus on German and on regulated domains such as public administration, industry, and aerospace. To support these settings, we control the full model-development process, including data, architecture, training infrastructure, post-training, and evaluation. We train Kolibri on 24T tokens across pre-training, mid-training, and long-context extension, with German accounting for more than 20% of the data mix by bytes. To reach this share, we complement German datasets with more than 2T German tokens that we curate from the web with our own pipeline and generate synthetically. Post-training combines supervised fine-tuning on curated German and English data with reinforcement learning on more than 1.2M internally curated tasks, which cover reasoning, tool use, instruction following, code, and retrieval. The model activates 4.4% of its parameters per token, and among the open models evaluated in this report, it lies on the Pareto frontier of quality and serving cost in English and German. Kolibri builds on Kolibri Origin, an earlier iteration that validated our full pipeline at scale. Both models come from our Model Factory, which connects data, experimentation, training, evaluation, and deployment through version-controlled workflows.

0.41

RankAlign: Vision-Language Representation Alignment via Rank Transformation

cs.CV ·2026-10-22 ·Enzhe Zhao, Marco Fiorucci, Lamberto Ballan
top venue
Abstract · Accepted at NeurIPS 2026

While the Platonic Representation Hypothesis posits that vision and language models converge toward compatible latent geometries, recovering class-level correspondences between unpaired modalities without cross-modal supervision remains challenging. Existing relational matching methods compare within-modality similarity kernels directly, although the same semantic structure may be represented through different numerical parameterizations across modalities. We propose RankAlign, a rank-based normalization strategy for class-level unpaired vision-language matching. RankAlign transforms each within-modality similarity kernel into an ordinal relational representation before permutation matching, while leaving the correspondence space and matching solver unchanged. The resulting representation is invariant to strictly increasing reparameterizations of similarity values while preserving the ordinal structure used for matching. We provide exact-recovery and stability guarantees that characterize correspondence recovery under both exact and partial ordinal agreement. Experiments across diverse frozen vision and language encoders and multiple datasets demonstrate improved and more stable class-level correspondence recovery, including under label corruption and domain shift.

0.41

Beyond Watching and Searching: Benchmarking Agents for Video-centered Deep Research Reports

cs.CV cs.CL ·2026-10-22 ·Jiahao Meng, Qize Yang, Yuxuan Wang, Yunfei Chu, Wensi Dai +2 more
code/data
Abstract

Deep research agents synthesize open-web evidence into long reports, but current benchmarks do not evaluate research centered on a source video. We introduce Video-Centered Deep Research Report Generation. Given a video and user request, an agent identifies research questions, retrieves external multimodal evidence, and produces an interactive report linked to relevant video moments. This task has direct practical value because it helps users locate relevant moments, verify claims, reproduce demonstrated procedures, and make informed decisions from video content. We present a skill-based workflow and OmniDR-Bench for this task. For evaluation, we propose Omni Atomic Rubric Trees (Omni-ARTs). Each tree contains three dimensions, 23 shared directions, and task-specific atomic binary criteria. The directions test audiovisual grounding, multimodal evidence use, and interactive presentation. Audited Rubric Construction (ARC) builds these trees through independent evidence collection, automated checks, behavioral calibration, and iterative human review. OmniDR-Bench contains 50 Chinese and English tasks across 19 realworld subcategories. It includes 6,414 criteria. Across 18 agent configurations, the best system scores 63.3 out of 100. Current agents still need broader content coverage, stronger causal and logical analysis across modalities, and more accurate citation support. We further analyze models, harnesses, skill configurations, video categories, and research trajectories. The code and data are released at https://github.com/marinero4972/OmniDR-Bench.

0.41

Bound-Conditioned Latent Inference for Progressive Image Compression

eess.IV cs.CV ·2026-10-22 ·Jaeseok Jang, Seungmin Jeon, Kwang Pyo Choi, Chang-Su Kim
top venue
Abstract · Accepted to NeurIPS 2026

Progressive image compression reconstructs images from prefixes of a single bitstream and therefore involves sequential latent inference under partial observations. Existing trit-plane codecs mainly treat decoded trits as discrete symbols for probability prediction, without fully exploiting their role as interval constraints on the underlying continuous latents. We propose bound-conditioned latent inference (BLI), a framework that reformulates progressive trit-plane coding as sequential latent inference under shrinking interval constraints. At each coding step, the trits decoded so far define lower and upper bounds for each latent element. BLI uses the center and width of these bounds, together with hyperprior information shared by the encoder and decoder, to jointly refine the Gaussian parameters $(μ,σ)$ for entropy modeling and the latent estimate $\hat{y}$ for reconstruction. As additional trits are decoded, the feasible intervals shrink monotonically, yielding tighter constraints for the next inference step and enabling flexible in-plane refinement schedules. Experiments on Kodak, CLIC, and JPEG-AI show that BLI achieves state-of-the-art rate-distortion performance among progressive image codecs, reducing BD-rate by $17.78\%$ over DPICT and by $7.96\%$ over CTC on Kodak, while replacing CTC's plane-specific predictors with a shared model that uses about $5.3\times$ fewer parameters and achieves lower latency than CTC.

0.41

Meaning Contracts: Local Semantic Geometry in Large Language Models

cs.CL cs.AI ·2026-10-22 ·Andrés Daniel Godoy Ortiz
code/data
Abstract · 24 pages, 8 figures. Code and data: https://github.com/adgodoyo/Meaning_Contracts

This paper studies the local geometry of semantic meaning in large language models. Current representation analyses ask whether meanings are close in embedding space, or whether linear probes separate semantic classes, but they do not characterize the within-concept local geometry: what is invariant across realizations of a single semantic factor, what varies, and what changes when meaning is refined. This paper characterizes that geometry through two controlled probes evaluated on four architecturally diverse decoder-only LLMs (Gemma 4, Llama 3.2, Qwen 2.5, Pythia). A cross-lingual probe uses language as a perturbation to separate exact meaning matches, same-factor realizations, nearby concepts, and out-of-domain contrasts. A refinement probe separates specificity-bearing semantic narrowing from high-probability attributive elaboration and adversarial controls; both probes are validated by blind human judgments. Four findings recur across all models: under mean pooling, once dominant variance directions are removed, sentence pairs order by semantic relatedness (same factor, then same domain, then different domain) without overlap over a band of layers; semantic factors organize around an invariant core surrounded by structured realization-specific variation; refinement produces directed displacement into tighter neighborhoods, including against controls built on the same base noun; and sentence-level and token-local readouts expose complementary geometry (volume contraction at the sentence level, sharper anisotropic endpoints at token-local readout). We interpret these results as evidence that LLM semantics are organized as local anisotropic neighborhoods in which meaning contracts under refinement.

0.41

Large-area Spatially Aligned Anchors

cs.MM cs.CV ·2026-10-22 ·Joni Vanherck, Brent Zoomers, Lode Jorissen, Isjtar Vandebroeck, Eric Joris +1 more
published
Abstract · Accepted at XR Salento 2024

Extended Reality (XR) technologies, including Virtual Reality (VR) and Augmented Reality (AR), offer immersive experiences merging digital content with the real world. Achieving precise spatial tracking over large areas is a critical challenge in XR development. This paper addresses the drift issue, caused by small errors accumulating over time leading to a discrepancy between the real and virtual worlds. Tackling this issue is crucial for co-located XR experiences where virtual and physical elements interact seamlessly. Building upon the locally accurate spatial anchors, we propose a solution that extends this accuracy to larger areas by exploiting an external, drift-corrected tracking method as a ground truth. During the preparation stage, anchors are placed inside the headset and inside the external tracking method simultaneously, yielding 3D-3D correspondences. Both anchor clouds, and thus tracking methods, are aligned using a suitable cloud registration method during the operational stage. Our method enhances user comfort and mobility by leveraging the headset's built-in tracking capabilities during the operational stage, allowing standalone functionality. Additionally, this method can be used with any XR headset that supports spatial anchors and with any drift-free external tracking method. Empirical evaluation demonstrates the system's effectiveness in aligning virtual content with the real world and expanding the accurate tracking area. In addition, the alignment is evaluated by comparing the camera poses of both tracking methods. This approach may benefit a wide range of industries and applications, including manufacturing and construction, education, and entertainment.

0.41

Can an Agent Debug a Web Corpus? Agentic Auditing of Slovak FineWeb2

cs.CL ·2026-10-22 ·Marek Šuppa, Ivan Vykopal, Jaroslav Kopčan
publishedhot direction
Abstract · Accepted to WaC Workshop at EMNLP 2026

Multilingual pipelines such as FineWeb2 apply one recipe across a thousand languages, leaving the failure modes of individual languages unanticipated. We use an LLM once, as a corpus debugger rather than a classifier: a tool-using agent explores the corpus through a read-only, budgeted API and compiles each finding into an executable test, which a native speaker then validates. On the 26.5M-document Slovak portion of FineWeb2, three runs costing \$2.21-\$2.84 each found seven findings over four phenomena. The largest is a machine-generated content farm of 1,408,437 documents (5.32%) that passed the pipeline's filters, 86.6% of it passing even stricter thresholds. Blinded labels support 6 of the 7 findings (defect rates .53-1.00 against .08 in a random control) and do not support one. The compiled suite scans the full corpus in 0.7 s of wall-clock time without a GPU, against \$12k-\$90k per LLM classification pass. The best static sampling baseline validates more selectors than any agent run, at under half the cost. The agent contributes exact prevalence, a precision estimate checked against sampled documents, an account of why the pipeline missed each slice, and a reusable test. We release the framework, Slovak test suite, validation data and agent trajectories.

0.41

Diagnostic Loss: Asymmetric Supervision for Radiology Report Generation

cs.CV cs.LG ·2026-10-22 ·Ignacy Kolton, Wenxuan Li, Zongwei Zhou, Jarosław B. Ćwikła, Arkadiusz Sitek +2 more
code/datahot direction
Abstract · 26 pages, 6 figures, 7 tables. Code available at https://github.com/gmum/DLoss

Automated radiology report generation requires models to do more than produce fluent clinical narratives: they must explicitly state supported diagnoses while avoiding unsupported ones. However, existing medical vision-language models (MedVLMs) are typically trained with token-level maximum likelihood objectives, which prioritize text agreement over diagnostically critical terms. As a result, models can miss key diagnoses and introduce wrong ones. We identify this as an objective misalignment between conventional language modeling and the asymmetric diagnostic requirements of radiology reporting. To encode these requirements, we introduce Diagnostic Loss (DLoss), a differentiable, architecture-agnostic objective that operates directly on decoder token probabilities. DLoss separately aggregates probability mass over vocabularies associated with diagnoses supported by the reference report and with unsupported alternatives. Using asymmetric pooling across token positions in the report, it encourages supported diagnoses to receive sufficient probability in at least one position while suppressing spurious probability spikes for unsupported diagnoses across all positions during training. The loss can be applied to existing 3D MedVLMs without modifying their vision backbones, language decoders, or inference procedures, and introduces no additional inference-time computation. We train and evaluate DLoss on 50,000 abdominal CT scans across several 3D MedVLM architectures on both in-distribution and out-of-distribution report-generation benchmarks. Across these settings, DLoss improves diagnostic faithfulness while maintaining competitive performance on conventional language-generation metrics, showing the value of asymmetric sequence-level supervision for radiology report generation. The code for DLoss is available at https://github.com/gmum/DLoss.

0.40

ReCoG: Relational Concept Graph for Training-Free Personalization

cs.CV ·2026-10-22 ·Changseon Yu, Hyeokjun Kweon
benchmarkpositioncode/data
Abstract

Training-free personalization adapts a frozen vision-language model (VLM) to recognize user-defined concepts from only a few exemplars by retrieving concept entries at inference time. While recent retriever-based methods enrich each concept with descriptive evidence, they largely treat concepts in isolation, so the retrieved information can be plausible yet non-discriminative, especially when the personalized concept set contains highly confusable, fine-grained categories. We argue that personalization is inherently relational: correct recognition depends on the specific distinctions between a concept and its nearest alternatives. To this end, we propose ReCoG, a Relational Concept Graph for retrieval-augmented personalization. ReCoG represents concepts as nodes and augments them with directed edges that explicitly capture discriminative cues between concept pairs, elicited by a frozen VLM during database construction. At inference, ReCoG retrieves a shortlist and resolves ambiguity via explicit pairwise comparisons conditioned on the corresponding relational edges, selecting the concept that consistently wins against competing candidates. We also introduce FINGER benchmark to test personalized recognition on confusable concept sets. Across FINGER, ReCoG substantially outperforms prior methods, with the largest gains in the most confusable regimes. ReCoG also achieves consistently best performance on established personalization benchmarks including captioning and VQA, confirming that relational evidence is broadly beneficial beyond fine-grained settings. Project Page: https://csy-98.github.io/ReCoG/

0.40

Zero-Shot Domain Generalization via Domain Prompt Learning

cs.CV ·2026-10-22 ·Shiming Chen, Shuhuang Chen, Salman Khan, Fahad Shahbaz Khan
top venuebenchmarkcode/data
Abstract · Accepted to IJCV

Domain generalization (DG) aims to improve the generalization of the model to any unseen domains. Targeting this goal, existing DG methods mainly focus on learning domain-invariant features or prompts from the given source domains. However, these methods rely on source-domain data, and collecting and annotating large-scale single- or multi-source datasets is often infeasible, limiting the practical applicability of DG. To tackle this challenge, we propose {\underline{Do}}main \underline{P}rompt Learning (DoP), a method built on a large vision-language model (e.g., CLIP) for zero-shot domain generalization, which recognizes unseen classes in unseen target domains without access to source-domain data. DoP essentially learns the target domain knowledge via prompt both in text and visual modalities to overcome the domain shift in text-visual modalities under the test-time adaptation formulation. Specifically, DoP first takes text domain learning (TDL) to simulate various distribution shifts by learning diverse style tokens implicitly in the text space with inter-style diversity loss and intra-style consistency loss. Subsequently, DoP employs visual domain learning (VDL) to further refine the target-domain-relevant style token generated by TDL by making use of the test visual data with self-supervised learning. Extensive experiments on popular DG benchmark datasets (e.g., PACS, VLCS, OfficeHome, and DomainNet) and ImageNet show that our DoP consistently outperforms the state-of-the-art DG approaches. Codes are available at: https://github.com/shiming-chen/DoP.

0.40

Mining Frames: Association Rule Mining as an Unsupervised Framework for News Framing Detection

cs.CL cs.CYcs.IR ·2026-10-22 ·Hager Hesham Mustafa
code/data
Abstract · 35 pages, 5 figures, 4 tables. Working paper. Dataset: https://www.kaggle.com/datasets/anmatproject/framings-gaza-arm-dataset

Computational framing analysis is dominated by supervised classifiers that require large annotated corpora and predefined frame taxonomies, limiting their use for non-English languages, novel events, and low-resource settings. This paper presents Association Rule Mining (ARM) as an unsupervised, annotation-free framework for news framing detection. Each document or sentence is treated as a transaction of lexical items, and high-lift co-occurrence rules are read as candidate framing signals (Gamson and Modigliani's condensing symbols), confirmed in context by journalist-researchers. The primary validation is Framing Gaza: 25,512 English-language articles from 16 US, UK, French, and German outlets covering the first year of the Gaza war (October 2023 to October 2024), analyzed with Apriori (mlxtend), platform-weighted support, POS-filtered n-grams, and lift-based scoring. ARM inductively surfaces interpretable patterns without predefined categories: a cross-platform military frame, {israeli, strike} to {kill}, appears as a closed rule in 14 of 16 outlets, and a month-by-month analysis recovers event-driven framing pivots (e.g., the April 2024 World Central Kitchen strike) with no prior event labeling. A systematic headline-body divergence is identified, with headlines carrying concentrated, high-lift editorial framing and body text reproducing institutional, procedural language. A methodological validation on the Seven Days project (1,575 Arabic articles on the 2026 Iran war) shows that FP-Growth enables sentence-level analysis with far lower memory overhead. Findings were validated through editorial uptake: two published investigative stories (Daraj Media; Saheeh Masr). Preprocessing code is released openly; the full pipeline is documented but withheld to preserve operational deployability.

0.39

Reading, Writing, Sharing, and Interacting with the World through Video

cs.CV ·2026-10-22 ·Shengqiong Wu, Meng Luo, Bobo Li, Ziwei Liu, Shuicheng Yan +3 more
surveycode/datacommunity pick
Abstract · 104 pages, 18 figures, 19 tables. Project page: https://vid-gen.github.io/video-world-model-survey

Video foundation models have advanced rapidly in understanding and generating visual sequences, while video world models seek to capture the underlying dynamics of the world they depict. Despite their different objectives, both concern the relationship between video observations and an evolving world. This survey adopts world modeling as a unifying lens for examining these developments. Viewing an observed video as a partial, lossy, and temporally ordered projection of a latent temporal world, we organize the literature around four complementary operations: reading infers world properties from observed video; writing synthesizes video from specified or learned world conditions; sharing integrates understanding and generation through common modeling mechanisms; and interacting predicts and updates world states under actions and evolving observations. This perspective distinguishes temporal coherence in observed or generated sequences from consistency in the underlying world state. Across tasks, representations, models, datasets, and benchmarks, we trace a shift in emphasis from clip-level recognition and perceptual fidelity toward persistent state, controllable evolution, physical and causal validity, and evaluation grounded in interaction and downstream decisions, and we distill the open challenges that follow.

0.39

Harm Is Behavioural, Moderation Is Lexical: How Language and Culture Decide Whom Automated Safety Systems Protect and Whom They Silence

cs.CL cs.AI ·2026-10-22 ·Gabriel Lars Sabadin, Dr Nicola Harding, Sarah Kingston
positioncode/datahot direction
Abstract · 18 pages, 2 figures. Code, simulations and per-case outputs: https://osf.io/v7aef/files/osfstorage?view_only=9449667241894b1ca7d336fc4b8259d…

Content warning: this paper quotes slurs and hateful statements from published test suites; the racial slur among them is masked. Automated systems decide at scale which messages are harmful, and they decide unevenly. Classifiers label African American English as offensive up to twice as often as other English, the accounts of prominent drag queens can score as more toxic than those of white nationalists, and safety behaviour that holds in English collapses in low-resource languages. We argue that over-flagging and under-protection share one cause: deciding harm from vocabulary and surface form rather than from what a message does. We organise the evidence into seven failure modes and argue, as testable hypotheses, that harmful behaviours (a request for secrecy, a demand for payment, an attempt to isolate someone) travel across languages better than the words that express them. We propose an evaluation protocol that makes the gap between the best- and worst-served language communities a headline result, and show that the naive gap is inflated under equal service and that its usual bootstrap interval is unreliable. Applied to five open systems in eleven languages, from a keyword lexicon to two current multilingual guard models, the protocol finds that none reaches native-speaker accuracy at separating harmful function from harmless form (51 of 55 system-language pairs fall short); that greater capability raises discrimination in nine languages while leaving Arabic and romanised Hindi near chance, so the gap between the best- and worst-served languages grows with capability; that one system both over-flags and under-protects the same language most; and that a language a system reads worst can look the most fairly treated when only false positives are reported.

0.39

ULME: Predictive Coupling for Mutually Reinforcing Embedding and Language Modeling

cs.CL cs.AI ·2026-10-22 ·Yang Liu, Zhiqiang Yuan, Shu-Dong Huang, Jiancheng Lv, Liwei Yang +3 more
publishedpositioncode/data
Abstract · AAAI 2027 Submmitted

Text embedding and text generation are two of the most useful capabilities of modern large language models, yet they are typically provided by separate specialized models (an embedder and a generator) trained with incompatible objectives and deployed side by side in retrieval-augmented generation and agent systems. Recent unified embedding-generation models such as GritLM share parameters between the two capabilities, but they either switch attention regimes between modes or optimize embedding and language modeling as parallel objectives that never supervise each other. We argue that these designs are only partially unified: the two capabilities coexist on the same weights, but do not shape each other. We propose ULME, a post-training framework that unifies embedding and language modeling on a causal decoder-only LLM through predictive coupling. ULME inserts a chain of learnable slots between the query and the answer, reads a semantic embedding and a reasoning embedding from two designated slots, and trains three objectives on a shared causal backbone: a contrastive loss anchoring the semantic slot to document representations, a language-modeling loss supervising answer generation, and a stop-gradient predictive loss that aligns the reasoning slot with a pooled summary of the answer, so that generation actively shapes what the embedding encodes. Across two model families (Qwen2.5, Qwen3) and eight scales spanning 0.5B-8B parameters, a single ULME checkpoint improves MTEB by 3.3-4.1 points over the strongest open-source baseline, tightens BRIGHT and AdvBench-IR simultaneously, and remains on par with a generation-only fine-tune on standard accuracy benchmarks (MMLU, GSM8K, MMLU-Pro, HumanEval), demonstrating that genuine unification yields mutually reinforcing rather than merely coexisting capabilities.

0.39

Uncovering the Blind Spots of OKS-mAP: A Unified Protocol for Occlusion, Joint-Angle, and Temporal Evaluation of 2D Pose Estimators

cs.CV ·2026-10-22 ·Yifan Fan
code/data
Abstract · 15 pages, 7 figures

The dominant metric for 2D human pose estimation, OKS-mAP, has three blind spots: it does not distinguish visible from occluded keypoints, does not measure joint-angle error, and is silent on the temporal stability of tracking-free per-frame inference. We propose a supplementary protocol with three modules: (M1) OKS-mAP stratified by ground-truth visibility, with Delta_occ = mAP(v=1) - mAP(full), quantifying how much the aggregate metric masks occlusion-specific differences; (M2) joint-angle error (RMSE/MAE/P95) plus knee-valgus detection consistency; (M3) frame-to-frame displacement, median-filter residual jitter, and jump rate, before and after OneEuro smoothing. We ran it on four datasets (COCO val2017, CrowdPose, OCHuman, PoseTrack18) and four models from two architecture families (YOLO11s-pose, an occlusion-domain fine-tuned variant, ViTPose-B/L), under a common detector-free oracle protocol and an open-source harness. Delta_occ is negative in every combination (-0.19 to -0.47): aggregate mAP overestimates occluded-scene performance in every tested combination. Joint-angle error is amplified in occluded layers; angle metrics reveal differences mAP cannot localize, e.g., the smaller occlusion-trained model significantly outperforming a larger transformer in occluded elbows. On PoseTrack18 val, raw jitter matched between the two YOLO models; pixel-domain OneEuro smoothing cut jump rates 9-10% at no matched-scope AP cost; residual jitter persists after detector removal; the transformer pair's jump rate is significantly lower; smoothing benefits replicate across architectures. Stratified mAP tracks full mAP closely (sign-consistency 1.0, 12/12), serving magnitude diagnosis rather than re-ranking; angle-metric difference vectors (especially knee) partially diverge, suggesting independent information. The protocol and harness are open-sourced, complementing rather than replacing OKS-mAP.

0.39

HERD: Harness-Evolving Recursive Distillation for Continual Learning

cs.CL cs.AIcs.LG ·2026-10-22 ·Xun Sun, Thanh Thi Nguyen, Thuy-Trang Vu
Abstract · 22 pages, 9 figures

Continual learning requires a model to keep acquiring new skills and knowledge without compromising its existing capabilities. Parametric methods suffer from catastrophic forgetting and loss of plasticity. Non-parametric methods, such as prompts, skills, and harnesses, leverage in-context learning but are limited by context length and lossy plain-text compression. To effectively combine their complementary strengths, we propose HERD (Harness-Evolving Recursive Distillation), a hybrid continual learning framework that learns efficiently in the harness and internalizes experience into cumulative capabilities via distillation. On the harness side, HERD continually refines its skills to capture reusable patterns, while retaining high-residual examples that the skills fail to capture in a bounded memory. On the parameter side, HERD performs drift-constrained distillation with dual anchors, substantially reducing forgetting over long recursive runs. On the 25-stage task stream of TRACE+17, under fixed memory and compute budgets, HERD matches the strongest parametric baseline in new-task acquisition while exhibiting the least forgetting and the strongest retention of general capabilities.

0.39

MVLoMa: Multi-View Local Feature Matching with Any Geometry Foundation Model

cs.CV ·2026-10-22 ·Xinyang Lu, Shengzhe Wei, Sheng Tang, Weiqi Zhang, Juntong Fang +1 more
code/data
Abstract · Project Page:https://kosmoresearch.github.io/MVLoMa/

Local feature matching is typically formulated for image pairs, although structure-from-motion and visual localization commonly match one reference image against many retrieved or covisible neighbours. Applying a pairwise matcher independently to these centre-neighbour edges duplicates reference-side computation and prevents the matcher from exploiting complementary multi-view geometric evidence from the remaining observations. To address these limitations, we introduce MVLoMa, a sparse star-graph matcher that extends LoMa from image pairs to multiple views. MVLoMa first uses a frozen multi-view geometry model to jointly describe all images in a star, then injects the resulting cross-view geometric context into local descriptors. It then matches the centre image to all neighbours in a single forward pass through per-view self-attention and bidirectional centre-neighbour cross-attention. For supervision, we derive positive correspondences only from depth-valid, mutually visible dense warps and explicitly penalize high-confidence matches that violate epipolar geometry. On 25-view star graphs from HyperSim, ScanNet++, Mip-NeRF 360, and Tanks and Temples, MVLoMa obtains the best macro-average relative-pose AUC among the compared methods at all three thresholds (58.65/85.41/90.99). Its matcher is 2.20 times faster than sequential evaluation of the 24 centre-neighbour edges. The same design supports four tested frozen geometry encoders -- DA3, $π^3$, VGGT, and VGGT-$Ω$ -- with DA3-Giant performing best on all reported encoder-ablation metrics. Controlled ablations further isolate the effects of geometry-aware descriptors, visibility filtering, geometric negative supervision, and joint multi-view inference.

0.39

Synthetic-to-Real Text-Based Person Anomaly Search via Multi-Model Retrieval Fusion and Cross-Encoder Reranking

cs.CV ·2026-10-22 ·Eugene Denteh, Blessing Agyei Kyem, Andrews Danyo, Joshua Kofi Asamoah, Richard Dzinyela +3 more
benchmark
Abstract

Text-based person anomaly search requires a model to retrieve individuals from a large image gallery using natural-language descriptions of both appearance and behavior. This task is challenging because model training relies on synthetic image-text pairs, whereas evaluation is performed on real surveillance images containing domain differences, near-duplicate frames, and many visually similar distractors. We present a multi-model retrieval and reranking framework designed to improve synthetic-to-real transfer. The framework first adapts Qwen3-VL-Embedding-8B to synthetic person descriptions using bidirectional contrastive learning, explicit negative sampling, false-negative masking, and domain randomization. It then combines the normalized scores of six complementary Qwen3-VL checkpoints with those of a frozen SigLIP2 model using four-view gallery encoding. The fused retrieval scores identify the strongest candidates, which are jointly evaluated with the query using Qwen3-VL-Reranker-8B. The cross-encoder and retrieval scores are subsequently blended to produce the final ranking. We also construct a distractor-matched validation protocol entirely from synthetic data to select checkpoint weights, fusion coefficients, candidate depths, and reranking parameters without accessing the official test images. The final system achieves 94.7815% mAP and 90.8493% R@1 under the official AI City Challenge 2026 Track 4 test setting, with a reported eighth-place finish on the public leaderboard.

0.38

CVP-BAA: Efficiently Adapting Segment Anything Model for Medical Image Segmentation

eess.IV ·2026-10-22 ·Mei Wang, Minghui Li, Libin Lan
code/data
Abstract · Submitted to CVIU, 35 pages, 8 tables, 8 figures. Our code is available at \href{https://github.com/WangMei-wm/CVP-BAA}{here}

Various task-specific medical segmentation models have demonstrated strong performance, their heavy reliance on extensive annotations and highly specialized architectures limits their versatility, scalability, and clinical applicability. Recently, adapting vision foundation models, particularly the segment anything model (SAM), has emerged as a promising alternative. However, prevalent parameter-efficient fine-tuning (PEFT) strategies, such as prompt tuning and adapter tuning, often lack an explicit mechanism to embed local structural priors into the long-range dependency modeling of vision transformers (ViT), rendering them insufficiently sensitive to fine-grained details. To address these limitations, we introduce CVP-BAA, a novel hybrid PEFT method that improves SAM's sensitivity to domain-specific textures and local details, leading to superior boundary delineation. The proposed CVP-BAA adopts two key ideas. First, it introduces a convolutional visual prompting (CVP) strategy that leverages convolutional features as structural priors, prepending them to the key and value projections within the multi-head self-attention layers of strategically selected ViT blocks. Through a dedicated prompt generation-and-injection module, this mechanism effectively bridges local spatial details with global context semantics. Second, a boundary-aware adapter (BAA) is designed as a parallel branch to the feed-forward network within the same selected blocks. By explicitly computing and amplifying high-frequency components, the BAA enhances feature representations along tissue boundaries and fine-grained lesion structures. Extensive experiments on five benchmarks across three imaging modalities demonstrate that CVP-BAA achieves superior segmentation performance, surpassing existing task-specific models and PEFT-based SAM adaptations in both quantitative metrics and qualitative results.

0.38

Padding to k: A Minimum Count in the Schema Makes Open-Weight Models Pad Their Extraction Lists

cs.CL ·2026-10-22 ·Abhishek Barali, Aayush Paudel
code/dataunusual
Abstract · 26 pages, 4 figures. Preregistered at https://osf.io/e4r9f. Code and data: https://github.com/AbhishekBarali/padding-to-k

Extraction pipelines ask language models for JSON that follows a schema, and the schema can set a minimum number of list entries. We ask what models do when the document holds fewer items than that minimum. Under a preregistered plan, we test eight open-weight models on 558 real documents from nine annotated datasets, with six more models on hosted services as a reference. Three findings stand out. First, models treat the count as a target: a minimum shown in the schema, but not enforced, raises the share of lists longer than the true count from 34.5% to 55.4%. Second, the padding is hard to catch, because most added entries are real text taken from the document; about half of them pass a check that each entry occurs in the source. Third, common fixes help only partly: a line saying fewer items are fine removes the effect on average but misses more true items, a count in prose ("up to k") also raises over-counting, an "N/A" option under an enforced minimum still leaves half the lists padded, and re-asking with the validation error makes padding far more common. We release the code, answer lists and run records at https://github.com/AbhishekBarali/padding-to-k

0.38

Latent Coherence: Do Frontier Language Models' Probability Judgments Obey the Law of Total Probability?

cs.CL ·2026-10-22 ·Peter Potash
code/data
Abstract · 15 pages

LawZero's Scientist AI trains a Predictor toward the Bayesian posterior, with an objective that penalizes inconsistent probabilities among latent explanations. We ask whether current frontier language models, with no such training, are already consistent in this sense zero-shot: do their probability judgments satisfy the law of total probability, $P(X)=\sum_z P(z)P(X|z)$, when the latent $z$ must be inferred from records in the context? In separate single-turn calls, models are asked for $P(X)$, for $P(z)$ over the latent's states, and for $P(X|z)$; a causal family applies the same test to $P(y|do(x))$ and the back-door adjustment. Every query has an exact ground-truth probability, and a subset of items is asked in three wordings to separate persistent incoherence from rewording noise. Across seven models, GPT-6.1 Sol and Claude Opus 5.5 are the most coherent (mean absolute residual 0.08 and 0.12 log-odds, about 2 percentage points). Claude Haiku 5.5, Gemini 3.8 Flash and DeepSeek V4-Pro show persistent incoherence (0.27-0.35 log-odds) that is mostly item-specific rather than a shared offset. Within every model, larger residuals go with larger errors against the truth (Spearman 0.40-0.62), so the residual, which needs no ground truth, flags unreliable answers. Reasoning length does not predict coherence, across models or within a model across rewordings of the same question. Simple consistency checks pass almost everywhere; the failures are specific to reasoning through a latent variable. Code and data: https://github.com/ppotash/latent-coherence

0.38

The Fusion Tax: A Diagnostic Analysis of When Hybrid Retrieval Hurts

cs.IR cs.CL ·2026-10-22 ·Aaditya Chauhan, Arun Agarwal
publishedhot direction
Abstract

Hybrid retrieval, which fuses a sparse lexical retriever with a dense neural one, is widely recommended in the software documentation surveyed in September 2026, usually with reciprocal rank fusion (RRF), on the strength of aggregate gains. We ask what fusion costs on individual queries, treating "do not fuse" as a first-class outcome, in the setting in which hybrid retrieval is deployed: a standard Anserini BM25, five dense encoders spanning 22M-596M parameters and three model generations, a learned-sparse (SPLADE) alternative, and a cross-encoder second stage, on seven BEIR datasets and three fusion functions. Even the best method, dense-weighted score interpolation, ranks relevant documents below the better of its two components on 13-48% of queries and below the deployable dense-only retriever on 8-21% (14-48% and 11-25% at a weight selected under leave-one-corpus-out validation), while beating dense-only in aggregate on every corpus with the primary encoder; the downside is significant under loss-averse risk-sensitive evaluation on six of the seven corpora. We call this loss the fusion tax. It is mostly a ranking effect that preserves recall in aggregate, it is most likely on the queries that only one retriever covers, and it persists across the encoder ladder and the SPLADE substitution. RRF carries the higher tax wherever the dense retriever is clearly stronger and only beats dense-only where the two components are balanced. Fusion failure is weakly predictable within a corpus but transfers unreliably across corpora, because the features that predict it change sign with the sparse-dense balance. A cross-encoder recovers part of the tax only where it is itself reliable. We distill these findings into guidance that keys the fusion decision to the measured sparse-dense balance of the target corpus.

0.38

Spectra: Multi-Scale Implicit Decomposition for Deepfake Detection

cs.CV cs.AI ·2026-10-22 ·Dhananjaya Jayasundara, Kartik Narayan, Vishal M. Patel
position
Abstract

Deepfakes increasingly undermine trust in digital media, yet existing detectors often struggle to generalize across unseen manipulation techniques, compression levels, and dataset shifts. Most approaches rely on RGB appearance, frequency-domain artifacts, or their combination, learning cues present in the final synthesized image. While effective in-distribution, such cues are frequently generator-specific and can diminish as synthesis models evolve. We argue that a complementary perspective can be obtained through implicit neural representations (INRs), which expose how an image is reconstructed rather than merely how it appears. By fitting an INR to an image and analyzing its internal activations, one can reveal reconstruction dynamics that are difficult to observe in the image domain alone and may capture more intrinsic differences between genuine and synthetic content. To this end, we propose Spectra, a deepfake detection framework that combines semantic visual representations with artifact-aware cues derived from INR reconstruction. Spectra extracts multi-level reconstruction heatmaps from image-specific INRs and jointly models them with semantic features through a bidirectional fusion mechanism. This enables direct interaction between image semantics and reconstruction artifacts, allowing the model to identify forgery cues that are less dependent on specific generation pipelines. Extensive experiments on multiple deepfake benchmarks demonstrate improved robustness and cross-dataset generalization, consistently outperforming detectors based solely on RGB appearance or semantic representations. Furthermore, we show that the proposed representation extends beyond deepfake detection to face presentation attack detection, highlighting its potential as a unified framework for facial attack detection.

0.38

SmolVLA-MoE: Sparse Mixture-of-Experts for Efficient Multi-Task Robotic Manipulation

cs.CV ·2026-10-22 ·Shiyang Zhang, Miao Yang, Zhengda Li, Yang Shao, Ruifang Zhai +2 more
publishedbenchmarkcode/data
Abstract · 16 pages, 5 figures, 4 tables; conference paper prepared for IROS2027

Vision-Language-Action (VLA) models have shown promising results in multi-task robotic manipulation by leveraging visual and language understanding for action generation. However, existing lightweight VLA models employ a single dense action expert to handle all tasks, which limits their multi-task capability under constrained computational budgets. To address this issue, we propose SmolVLA-MoE, a novel framework that introduces sparse Mixture-of-Experts (MoE) into the FFN layers of the action expert in VLA models. SmolVLA-MoE replaces dense FFN layers with multiple parallel expert FFNs and a token-level router, enabling each token to adaptively select the most suitable expert for action generation. We further introduce a shared expert with a tunable weight parameter to balance the inheritance of pre-trained capabilities and task-specific expert specialization. Through sparse activation, SmolVLA-MoE activates only a single expert FFN during inference, with a total activated parameter count of approximately 500M including the frozen VLM backbone, while benefiting from approximately 800M total stored parameters. We evaluate our method on four task suites of the LIBERO benchmark. SmolVLA-MoE achieves an average success rate of 84.25%, outperforming SmolVLA (82.75%) and OpenVLA-7B (76.5%), while maintaining comparable inference efficiency. Our code is publicly available at https://github.com/dudu435/smolvla-moe.

0.38

Hierarchical Fine-Grained Cross-Modal Alignment via Dynamic Granularity Discovery and Optimal Transport

cs.CV cs.CL ·2026-10-22 ·Pengfeng Song, Shaozhong Zhang, Pengfei Shao
code/data
Abstract · 8 figures, 6 tables. Code available at https://github.com/sungkites/HFCMA

Fine-grained image-text retrieval requires a model not only to assess global semantic consistency between an image and a text, but also to model visual evidence for local textual semantics, including objects, attributes, quantities, actions, and relations, so as to distinguish semantically similar candidates. Existing methods commonly use fixed image patches, detected regions, or text tokens as the basic matching units and then aggregate local interactions through max pooling, cross attention, or similarity accumulation. Such strategies remain limited in semantic unit construction and global matching consistency. This paper proposes a hierarchical fine-grained cross-modal alignment framework, termed HFCMA. The framework progressively transforms low-level patch and token representations into adaptive visual regions and textual phrases, and then performs structured region-phrase matching within a global retrieval scoring process. Specifically, HFCMA first adopts dynamic granularity discovery (DGD) to aggregate raw patches and tokens into semantically compact visual regions and textual phrases. It then formulates region-phrase matching as an entropy-regularized optimal transport problem, thereby estimating a structured alignment matrix under global marginal constraints. To reduce background noise and local semantic drift, HFCMA further introduces background debiasing, textual importance weighting, and prototype consistency regularization. Experiments on Flickr30K and MS-COCO show that HFCMA effectively combines dynamic semantic unit construction with OT-based structured matching under a dual-encoder retrieval framework. The resulting fine-grained score provides structured local alignment information for candidate ranking and improves the discrimination of subtle semantic differences.

0.37

Pos4CHD: A Positive-Free Mask-Guided Counterfactual Synthesis Framework for Fetal Congenital Heart Defect Analysis

cs.CV ·2026-10-22 ·Wenjie Xuan, Yapeng Li, Hang Su, Chen Cheng, Juhua Liu +3 more
code/data
Abstract · Submitted to IEEE TMI, under review

The scarcity of positive cases remains a major bottleneck in developing automated diagnostic systems for fetal Congenital Heart Defects~(CHDs), particularly for rare subtypes. Although normal cases are accessible with effort, positive cases are hard to collect due to low incidence rates and privacy concerns, causing severe class imbalance or absence of positive training samples. Existing works resort to synthesizing positives with conditional generative models, but they are task-specific and require positive training samples, hindering application for fetal CHDs. To this end, we propose a positive-free mask-guided counterfactual generation framework for fetal echocardiography, namely Pos4CHD, which enables realistic, high-quality synthesis of pseudo-positive samples from semantic masks, with only normal samples at hand. Technically, we first develop FE-Gen, a mask-guided generative model tailored for fetal heart ultrasound synthesis that mitigates histogram drift and enhances image-mask spatial alignment. Building on FE-Gen, Pos4CHD employs a mask-to-image generation paradigm for CHD synthesis. It first manipulates semantic masks to simulate fetal-heart structural anomalies and then renders anatomically consistent images. Extensive experiments on in-house datasets, together with expert evaluations, demonstrate the visual fidelity~($-12.97$ FID) and anatomical consistency~($+3.96\%$ DSC) of synthesized images. Further validations across diverse real-world CHD diagnosis tasks exhibit promising results and consistent increments, including zero-shot classification~(achieving $93.25\%$ ACC on EbA and $72.92\%$ ACC on TAPVC), lesion detection~($+6.95\%$ mAP50 on VSD), and anatomy segmentation~($+10.49\%$ DSC on fetal heart) compared with baselines. Codes and models are publicly available on \href{https://github.com/DREAMXFAR/Pos4CHD}{GitHub}.

0.37

Simulated transcranial ultrasound of human language cortex: verifying heard speech needs far less data and signal than decoding it

eess.IV q-bio.NC ·2026-10-22 ·Omer Shalev
code/dataunusual
Abstract · 4 pages + references, 3 figures. Code: https://github.com/deShal3v/verify-vs-decode-ultrasound

Functional ultrasound could image cortical blood volume through the temporal bone, but the skull leaves a weak signal, and task-evoked cortical imaging through the intact adult skull has not yet been shown. We built a voxel-specific Monte Carlo link-budget simulation of a probe on the left temple and used it to degrade public fMRI of three people who heard 15 hours of stories. On the same anatomical voxels we trained a published state-of-the-art fMRI language decoder (Tang et al., 2023) and two verifiers that only check a candidate text. In clean data the verifiers needed far less training (11 to 12 minutes against 3.2 to 6.6 hours), but all three learned at a similar rate. Noise changed this qualitatively: the verifiers kept gaining evidence while the decoder slowed and, under simulated transcranial physics, stalled, whereas the verifiers still reached their goal under a favorable simulated link budget. This asymmetry suggests that low-SNR modalities such as transcranial ultrasound may support verification well before decoding.

0.36

HiP3R: Hierarchical Pose Graphs for Long-Horizon 3D Reconstruction

cs.CV ·2026-10-22 ·Zunnan Xu, Shangchen Zhou, Yihang Luo, Chen Change Loy
community pick
Abstract · Technical Report

Feed-forward models accurately estimate camera motion and 3D geometry from short video windows. However, computational and memory costs prevent direct inference on sequences containing thousands of frames. Long videos must therefore be processed as overlapping chunks. This division produces independent local reconstructions with inconsistent coordinate frames and duplicate estimates of shared cameras. Achieving global consistency requires propagating constraints from adjacent overlaps and distant revisits across the sequence. This requirement naturally motivates pose graph optimization. However, the choice of graph level creates a trade-off. Chunk-level graphs efficiently resolve scale drift and close large loops. Yet, their rigid transformations cannot correct internal errors or conflicting frame predictions. Conversely, frame-level optimization offers local flexibility, but scaling it globally is computationally expensive and can break metric scale. HiP3R resolves this trade-off using a hierarchical pose graph. HiP3R first optimizes a robust chunk-level $\mathrm{Sim}(3)$ graph to establish the global layout and scale. It then holds this layout fixed and estimates one shared $\mathrm{SE}(3)$ camera per frame, turning overlapping window predictions into a single trajectory. Experiments across multiple benchmarks demonstrate the effectiveness of HiP3R in improving camera trajectories and 3D reconstruction. Our method applies to both chunk-based and streaming backends without additional training or fine-tuning. The codes are available at https://kkakkkka.github.io/HiP3R.

0.36

Parameter-Efficient Adaptation of Large Language Models to a Low-Resource Agglutinative Language: A Corpus, Datasets, and a Reproducible Benchmark for the Bashkir Language

cs.CL ·2026-10-22 ·Mullosharaf K. Arabov, Svetlana S. Khaybullina
benchmarkhot direction
Abstract · Preprint. 31 pages, 1 figure, 17 tables

Bashkir is a low-resource agglutinative language written in Cyrillic and remains poorly represented in modern LLMs. We present an open resource suite and a systematic benchmark of adaptation methods: the Bashkir Web Corpus (71,567 documents, 46.9M tokens, 16 sources) and three labelled datasets for binary, multiclass and multilabel text classification; the benchmark covers the first two. Experiments use eight architectures (82M-8B: DistilGPT2, GPT-2 base/medium, Phi-3-mini, Mistral-7B, YandexGPT-5-Lite-8B, mBERT, XLM-R) and three adaptation groups (full fine-tuning, QLoRA, in-context learning), plus continual pretraining (CPT). (1) QLoRA on 3.8-8B models outperforms encoder-only fine-tuning on binary classification: YandexGPT-5-Lite-8B with CPT reaches macro-F1 0.748 vs 0.653 (XLM-R) and 0.650 (mBERT); pairwise differences are significant after Holm correction (p<0.001). (2) YandexGPT-5-Lite-8B significantly outperforms Mistral-7B and Phi-3-mini on multiclass classification, BPB and native-speaker ratings (mean 3.28 vs 2.52 and 2.22; n=100; alpha=0.69-0.77; p<0.05); on binary classification its QLoRA advantage is not significant (p>=0.084). (3) CPT effects depend on model and task: perplexity improves for Phi-3-mini (+0.58) and YandexGPT-5-Lite-8B (+0.33) but worsens for Mistral-7B (-1.15, p<0.001); on binary classification CPT helps Mistral-7B most (+4.0 macro-F1). (4) Tokeniser fertility on Bashkir varies 2.67x (2.92-7.80) and tracks adaptation quality, while perplexity across tokenisers is not comparable. Cyrillic tokenisers with Russian pretraining are preferable; LLM judges reproduce native ratings for grammar and coherence (rho=0.33-0.46) but not language correctness (rho=0.16, p=0.11); native evaluation remains necessary. All artefacts are open-licensed.

0.36

POS-Signature Demonstration Retrieval for In-Context Dependency Parsing

cs.CL ·2026-10-22 ·Muhammad Azam Afridi, Rajab Ali, Marco Passarotti
code/data
Abstract · 14 pages, 1 figure, 8 tables. Code and research artifacts: https://github.com/azamafridi23/pos-signature-parsing

Demonstrations for in-context learning are often retrieved by semantic similarity, which may be a poor proxy for relevance in structured prediction. We study demonstration selection for Universal Dependencies parsing with a POS-signature policy that represents sentences with weighted part-of-speech statistics and retrieves three length-compatible, gold-parsed examples from the labeled training split. We evaluate two open-weight language models without fine-tuning on all 2,077 sentences of the UD English Web Treebank test set, comparing the policy with shot- and format-matched fixed and semantic three-shot baselines, as well as zero-shot, chain-of-thought, and critique-refinement systems. POS-signature retrieval yields the highest labeled attachment score among the tested prompting systems for both models (72.4 and 73.2), outperforming semantic retrieval by 1.9 and 2.7 points; paired-bootstrap 95% confidence intervals exclude zero for both contrasts. Chain-of-thought and refinement comparisons vary across models. These exploratory results favor the tested POS-signature retrieval approach on English-EWT. We release the prompts, archived outputs, and evaluation code.

0.36

BrickWright: Agentic Creation of Physically Feasible Brick Builds

cs.CV ·2026-10-22 ·Ava Pun, Kevin Dai, Sally Chen, Maxwell Jones, Ruixuan Liu +5 more
positioncode/data
Abstract

Frontier coding agents are becoming increasingly adept at 3D modeling. Yet, they still face challenges when building objects meant to be assembled in the physical world, often creating invalid assemblies with floating, colliding, or disconnected parts. We argue that providing agents with tools for assembling builds and verifying their physical feasibility is sufficient for them to produce physically valid assemblies that still adhere to input prompts. Focusing on the domain of LEGO brick builds, we present BrickWright, a framework for generating physically feasible brick builds from text or image prompts using pre-trained agents, where a build is physically feasible if it uses only real parts, has no colliding or floating parts, and forms a connected structure if possible. BrickWright guides agents to create builds by writing code with BrickLayer, a Python toolkit with two main components: an ergonomic brick assembly API that halves inference cost compared to working in raw LDraw, and a validator that provides feedback on physical feasibility. Whereas baseline agents achieve physical validity rates ranging from only 9% (GPT-6 Luna) to 69% (GPT-6 Astra), BrickWright-guided agents achieve 100% validity on a sample of 512 prompts, while improving prompt adherence for lightweight agents and maintaining it for stronger agents. Additionally, they outperform specialized models on the text-to-brick task. Code is available on our project page: https://avalovelace1.github.io/brickwright/

0.36

MoG-VLM: Efficient Video Anomaly Detection through Motion-Guided Vision-Language Models

cs.CV cs.AIcs.LG ·2026-10-22 ·Khaled Waleed Dawoud, Muhammad Zaigham Zaheer, Khan Muhammad, Muhammad Haris Khan
publishedcode/data
Abstract · Accepted at BMVC 2026. Includes supplementary material

Vision-Language Models (VLMs) have shown strong potential for video anomaly detection, but dense frame-level inference makes them impractical for real-time deployment. Existing acceleration commonly relies on uniform sampling, which ignores content dynamics and can either waste computation on static segments or miss localized anomalies. We propose MoG-VLM a motion-adaptive anomaly detection framework that leverages motion purely as a lightweight control signal to decide when VLM inference is necessary. To this end, we introduce Temporal Gradient Magnitude (TGM) which performs content-aware sampling to concentrate computation on informative regions while avoiding redundant processing. We further introduce a hierarchical dual-stream design in which dense multi-frame VLM reasoning is activated only around frames with elevated anomaly likelihood, via anchor-triggered context-window expansion with early termination. A fusion module then reconciles these instantaneous semantic cues with temporal action dynamics. Across four benchmarks, our method achieves 91.91%, 84.30%, 87.88%, and 78.96% AUC-ROC on XD-Violence, UCF-Crime, MSAD, and ShanghaiTech, respectively, establishing new state of the art on three benchmarks under the zero-shot, training-free setting. On XD-Violence, we improve over the previous best result by +0.57% while delivering a 1.31x inference speedup. These results demonstrate that content-aware allocation of VLM capacity, processing fewer frames more intelligently rather than all frames uniformly, is a more effective axis of progress than uniformly densifying VLM inference across frames. Code is available on GitHub: https://github.com/kdawoud91/MoG-VLM.

0.35

Stream4D: Streaming 4D Reconstruction with Training-free Motion Decoupling via Causal KV-Cache Attention

cs.CV ·2026-10-22 ·Wei Zhang, Ying Feng, Mengtong Wu, Shiqiang Gong, Huaqi Zhang +1 more
top venue
Abstract · 10 pages, 8 figures. Accepted to ACM Multimedia 2026

4D scene reconstruction requires robust handling of dynamic objects, yet existing streaming methods commonly assume static scenes, leading to pose drift and ghosting artifacts over long sequences. We introduce Stream4D, a training-free, motion-decoupled streaming reconstruction framework that can be directly integrated into causal Transformer backbones (e.g., StreamVGGT, Stream3R). Our key observation is that causal attention implicitly encodes rich temporal motion cues within the KV cache. We adaptively aggregate these cues from multiple attention layers and across temporal windows, and fuse them with depth priors to handle challenging cases such as occlusions and textureless regions, producing precise dynamic masks that decouple static and dynamic regions. We then apply a soft key masking scheme that leverages these dynamic masks to suppress interference from dynamic regions in causal attention, thereby improving camera pose estimation accuracy. Stream4D inherits the efficiency of causal attention and remains compatible with existing streaming architectures. On diverse dynamic-scene benchmarks, Stream4D significantly improves camera pose estimation and static-scene reconstruction quality, yields reliable dynamic masks, and effectively mitigates long-sequence ghosting.

0.35

AgentVQA: A Unified Benchmark for Agentic Visual Understanding

cs.CV ·2026-10-22 ·Advait Gupta, Sachin Konan, Tom Yin, Gabriel Sarch, Zhuang Liu
benchmarkcode/data
Abstract · Project page: https://zlab-princeton.github.io/AgentVQA

Vision-language models (VLMs) can perform a broad range of tasks across diverse settings, yet their performance in agentic contexts remains poorly understood. Although online evaluations in simulators and real-world environments are essential for measuring interactive agent performance, they are costly and challenging to scale. Complementary offline VQA benchmarks can enable faster and more reliable comparisons across agentic domains, yet existing benchmarks remain limited and fragmented. To address this gap, we introduce AgentVQA, an offline benchmark for evaluating VLMs on realistic agentic scenarios across five agent domains: Web Agents, Robotics, Egocentric Videos, Games, and Spatial Planning. We build AgentVQA by manually selecting 1000 challenging, real-world agent examples from 23 source datasets and adapting them for reliable offline evaluation. The resulting benchmark preserves realistic scenarios from interactive environments across diverse domains, and its model rankings track those from corresponding online evaluations. Across the 22 evaluated models, the best achieves only 58.5% accuracy. Our cross-domain analysis shows that errors are driven primarily by failures to ground and maintain accurate representations of the current and evolving visual state, including localizing relevant UI elements, interpreting motion, tracking state changes over time, and inferring 3D spatial relationships, rather than by task misinterpretation or downstream logical reasoning.

0.35

Would an LLM Match into a Residency? Uncertainty, Reasoning and Image Use of Open Language Models on the Spanish MIR Examination

cs.CL cs.AIcs.CYcs.LG ·2026-10-22 ·Miguel Fernandez-de-Retana, Unai Zulaika, Ruben Sanchez-Corcuera, Aritz Bilbao-Jayo, Aitor Almeida
hot direction
Abstract · 13 pages, 8 figures

Accuracy on medical examinations is common evidence offered for the clinical use of language models, yet a health system that intends to run an open model on its own hardware needs to know considerably more: whether the model recognizes when it is likely to be wrong, whether medical fine-tuning improves this ability, whether its correct answers rest on sound clinical reasoning, or whether it actually uses the images it is given. We examine these questions for four open models of 8 to 27 billion parameters on the Spanish MIR residency examination, scoring 1,552 questions from eight editions under the exam's own rules. Answering every question, the three larger models would rank among roughly the first 600 to 3,000 candidates, with 41 to 43 specialties still open to them. Abstention, however, never pays, because even their least confident answers are right more often than the 25% break-even. Medical fine-tuning reduced overconfidence at the cost of accuracy, without improving how well confidence separates right from wrong answers. Moreover, 13-38% of correct answers rejected a distractor for a clinically unsound reason that confidence did not reflect, and removing or swapping the provided contextual images left accuracy unchanged.

0.34

GM-GReFEL: Geometry-Guided Spatiotemporal State-Space Learning for Robust Dynamic Facial Expression Recognition

cs.CV cs.LG ·2026-10-22 ·Yudhistira Arditya Pratama, Yi-Zeng Hsieh
code/data
Abstract

Recognizing facial expressions in unconstrained video suffers from two coupled problems: self-attention grows quadratically in cost with clip length, and crowdsourced labels are noisy. We present GM-GReFEL, a spatiotemporal state-space network with a geometry-aware decision stage. Its Modulated GroupMamba backbone replaces a spatiotemporal Vision Transformer. Six channel groups are scanned separately, four over the frame plane and two over time, so the cost grows linearly with sequence length ($\mathcal{O}(N)$); after each layer, a Channel Affinity Modulation gate mixes the groups again. On top of the backbone, a GReFEL reliability module turns the normalized entropy of each prediction into a mixing weight between the classifier output and a vote from trainable geometric anchors, so that facial geometry counts for more when the network is unsure. The model is initialized by static-to-dynamic weight transfer and trained with a Triple Loss. In five-fold cross-validation, the best fold scores 66.57\%/76.46\% UAR/WAR on DFEW, 0.3 points or less below the strongest compared method, and 45.66\%/62.27\% on MAFW, ahead of all compared methods; on FERV39k the mean over five splits, 42.30\%/52.66\%, is competitive. The five-fold means are 64.07\%/74.54\% on DFEW, on par with recent large pre-trained models, and 38.90\%/54.37\% on MAFW, below the best of them. A paired $t$-test on DFEW indicates a modest benefit of temporal modeling (+1.61 UAR points, one-sided $p=0.031$). Attention stays on plausible facial regions, and latency is close to a ViT-B/16 baseline (47.4 vs.\ 46.8 ms per clip). Code: https://github.com/ardiytama/GroupMamba-DFER.

0.34

Delta N-gram: Forward-Only Continual Writing into a Language Model's Native N-gram Memory

cs.CL cs.LG ·2026-10-22 ·Hao Wu
code/data
Abstract · 33 pages, 3 figures, 9 tables, 2 algorithms. Follow-up to arXiv:2610.14921. Code: https://github.com/af-74413592/delta-ngram

Our previous analysis found that a model's native hashed N-gram table is where continual updates disturb the base least, but its writes needed backpropagation and changed the pretrained table. We present Delta N-gram, a forward-only write path into that table. The pretrained table T^0 is never modified. Evidence is written into a sparse patch Δon the rows it addresses: a 4.1M-parameter writer emits a target and a gate per (position, hash head) event, and a block-averaged delta rule keeps each row a convex combination of its old value and the gated targets. The writer is trained offline on raw text with the pretraining objective; at use time writing takes forward passes only, and clearing Δrestores the original model. A 0.6B prototype trained to answer questions keeps 16/16 facts after a second write; online gradient writes keep 2--4/16. We study two writers: a plain writer that encodes evidence with the T^0-only model, and a delta writer that reads its current memory while encoding. On Qwen3.8-Flash-Next, the plain writer's memory is document-specific: on 64 held-out news events it lowers continuation NLL by 0.068 nats/token against empty memory and by 0.024 against another document's memory, and 0.063 and 0.013 after 16 intervening articles. The delta writer's memory is more document-specific and keeps more of it under distraction (70% vs 54%). Over 35,000 writes on our previous seven-task chain both writers stay stable and the delta writer drifts less (layer-2 math drift 176% vs 188%); later accuracy changes by -0.26/-0.52pp (-7.29pp for gradient updates) and small-k_i drift is 17.9/17.5% (39.2%). Acquisition is negative, and general-input drift reaches 32.8% (20.75%): format rows read by every example grow to 4--12 times their pretrained norm. Zeroing eight such rows removes over 98% of layer-2 drift on math inputs, together with their NLL gain.

0.34

Diffusion-Based Synthesis of Complete 3D Biparametric Prostate MRI

cs.CV ·2026-10-22 ·Claudia Giardina, Montse Pardàs, Verónica Vilaplana
publishedhot direction
Abstract · Presented at the 6th Deep Generative Models Workshop (DGM4MICCAI), MICCAI 2026

Synthetic biparametric MRI (bpMRI) may help alleviate the limited availability of large and representative prostate MRI datasets for deep learning applications. In this work, we present a conditional 3D diffusion framework for anatomy-conditioned synthesis of complete prostate bpMRI from anatomical masks. The framework supports sequential synthesis, in which T2w and diffusion modalities are generated by separately trained models, and joint synthesis, in which all bpMRI modalities are generated simultaneously within a single model. Five configurations were trained using PI-CAI, FastMRI Prostate, and PROSTATEx datasets to investigate different conditioning schemes, target modalities, and synthesis strategies. Evaluation using distribution-based metrics, paired-image similarity, and qualitative assessment demonstrated the feasibility of both synthesis pathways. Joint synthesis achieved competitive distributional fidelity while generating complete bpMRI, whereas direct synthesis of ADC and computed high-b-value DWI outperformed their derivation from synthetic diffusion-weighted images.

0.34

Residual Calibration via Local Feature-Space Refinement

cs.CV cs.LGstat.ML ·2026-10-22 ·Xihao Wang, Chengyao Yu, Ruixing Ming
top venue
Abstract · Accepted at NeurIPS 2026

Deep neural networks often struggle to provide reliable confidence estimates, making post-hoc calibration essential for reliable decision-making. Existing post-hoc methods largely rely on score-based transformations, which improve marginal calibration but often fail to capture spatially varying miscalibration in local regions of the learned feature space. To address this gap, we propose Local Adaptive Probability Estimation (LAPE), a post-hoc residual calibration framework. Specifically, LAPE treats the global calibrator output as a stable anchor and refines the individual prediction by leveraging non-parametric local evidence from neighboring calibration samples. As a plug-and-play refinement module, LAPE effectively utilizes the information about feature-space geometry and can reuse the calibration data that are used to construct the global calibrator. Furthermore, we extend LAPE to the setting with multi-view inputs through Cascade Feature Augmentation Fusion (CFAF). Extensive experiments demonstrate that the proposed methods can improve the calibration quality of various existing baselines and remain efficient and robust across different settings.

0.34

HARMONY: Hierarchical Anchor Retrieval on Manifold for Oblivious-source Machinery Acoustic Anomaly Detection

cs.SD ·2026-10-22 ·Yuhang Zhang, Yang Li, Qiang Wu, Mingyuan Li
positioncode/data
Abstract

In standard acoustic anomaly detection (ASD), models are usually trained separately for each machine or domain using explicit metadata. However, in realistic deployments, machine identifiers are incomplete, unreliable, or simply unavailable. Thus, metadata-free machinery ASD asks a single model to monitor mixed machine populations without machine IDs. In this paper, we argue that the main failure mode is not a weak detector family, but representation entanglement. The heterogeneous normal modes overlap in frozen pre-trained feature spaces, making anomaly scoring and approximate retrieval unreliable. We study this problem directly and propose HARMONY (Hierarchical Anchor Retrieval on Manifold for Oblivious-source acoustic aNomalY detection), an anchor-guided geometric repartitioning framework that reorganizes mixed-source features into compact Voronoi regions and reuses this structure for hierarchical retrieval. On DCASE 2020, HARMONY improves mean AUC from 75.40% to 90.00% (14.60 percentage points) over the strongest unified baseline and from 83.90% to 90.00% over a strong foundation-feature baseline. On a single CPU, its hierarchical retrieval achieves a 7.8x speedup over exhaustive search with only a 0.20 percentage-point AUC drop. These results suggest that incorporating explicit geometric partitioning can improve both detection accuracy and retrieval efficiency for ASD. Code is available at https://github.com/Zzyh0208/HARMONY.

0.34

WeVisOne-VL: Unified Visual Perception from Objects to Pixels

cs.CV ·2026-10-22 ·Tianming Liang, Junyi Pan, Yuxi Li, Canmiao Fu, Zhipeng Huang +3 more
code/datacommunity pick
Abstract

Existing vision-language models (VLMs) remain unreliable for fine-grained visual perception. In this report, we present WeVisOne-VL (WeChat Vision All-in-One), a powerful VLM that unifies object-level and pixel-level perception through structured next-token prediction. It supports multiple visual tasks including object detection, referring localization, counting, OCR, segmentation, and matting, enabling localization at any user-specific granularity, from points and boxes to masks and alpha mattes. Architecturally, we integrate a lightweight pixel decoder into the VLM via a special <OBJ> token to generate instance-level pixel outputs. We train the model with a three-stage curriculum SFT followed by mixed-task RL. Across over 30 benchmarks, WeVisOne-VL demonstrates leading performance on both object-level and pixel-level tasks, while maintaining robust rejection for object-absent queries. By releasing WeVisOne-VL to the open-source community, we hope to advance VLMs from broad visual understanding toward truly reliable perception systems for real-world deployment.

0.34

Attention-guided classification and tumor localization of combined hepatocellular-cholangiocarcinoma

cs.CV ·2026-10-22 ·Xiaowen Liang, Hélène Cazier, Aurélie Beaufrère, Miguel Albuquerque, Valérie Paradis +1 more
Abstract

Combined hepatocellular-cholangiocarcinomas (cHCC-CCA) are a type of cancer that share components of two other primary liver cancers: hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCA). Accurate diagnosis of cHCC-CCA, of which the behavior may resemble that of either HCC or CCA, remains a challenge due to strong intratumoral heterogeneity. Since the prognoses differ between liver cancer types, it is necessary to improve their morphological characterization and recognition in order to accurately identify cHCC-CCA. In this work, we use whole slide images (WSIs) of the two main types of liver cancer (HCC and CCA) and cHCC-CCA, stained with Hematoxylin-Eosin-Safran (HES), to train a classification model that can distinguish between these subtypes and provides a potential solution for localizing different types of tumors in the combined liver cancer images.

0.33

Where Do Long Captions Fail? Position-Aware Diagnosis and Reinforcement Learning for Detailed Image Captioning

cs.CV ·2026-10-22 ·Yuanze Hu, Zhichao Yang, Junwei Jing, Xin Yang, Wenxuan Liu +7 more
top venue
Abstract · Accepted at NeurIPS 2026. 40 pages, 3 figures; includes references, appendices, and the paper checklist. Updated figures, captions, and appe…

Detailed image captioning is usually framed as an amount problem: a model should mention more visual facts while avoiding hallucination. However, autoregressive VLMs generate captions as ordered sequences, and later generations may be produced under weaker or noisier visual grounding. We therefore argue that long-form captioning should be studied not only by what facts are mentioned, but also by where they appear. We introduce a position-aware rubric representation that decomposes dense human references into atomic visual claims, verifies generated captions against those claims, and induces a quartile-wise composition of supported, hallucinated, and generic content. This lens reveals two qualitatively different failures in long captions: models tend to produce unsupported specificity in the middle, while the final portion often escapes into vague, non-evidential language rather than simply accumulating more hallucination. A same-reference human control does not exhibit these patterns; additional human audits examine the visual validity of the induced labels. Motivated by this diagnosis, we propose PAGER (Position-Aware Grounded Evidence Reinforcement), a GRPO training objective with separate signals for supported fine-grained middle evidence, evidence-bearing tail continuation, claim coverage, and content-aware length control. The result is a closed loop: the same ordered claim representation identifies where long captions fail and supplies the position-specific reward needed to repair those failures. Experiments on detailed-caption benchmarks show that PAGER improves position-wise grounding and pairwise caption preference while preserving competitive global caption quality.

0.33

LACToS: Local Anomaly Detection via Conditional Typicality of Superpixels

cs.CV cs.LG ·2026-10-22 ·Jihoon Hong, Alireza Kheirandish, Rohan Sanda, Sara Fridovich-Keil
code/data
Abstract

Local anomaly detection is the task of identifying spatially confined anomalous or out-of-distribution (OOD) regions within an otherwise in-distribution (ID) image. Most existing methods are designed for a specific context, with a strong focus on industrial defect detection, where they achieve excellent performance on well-established benchmarks. However, we observe that these methods can exhibit significant performance drops in other contexts, especially when the inlier distribution is multimodal. To account for this inlier complexity, we propose LACToS, a principled local OOD detection method based on statistical typicality and made tractable through two innovations: (i) grouping pixels into superpixels and (ii) using a computationally efficient proxy to estimate typicality. We evaluate LACToS across three diverse applications spanning medical imaging, human faces, and X-ray security screening. LACToS performs more robustly than prior methods across these benchmarks. The code is released here: https://github.com/voilalab/LACToS.

0.33

DirCast: Live Esports Directing and Commentary with a Single Streaming Video-Language Model

cs.CV ·2026-10-22 ·Zixian Gao, Lixin Xiu, Penghao Luo, Xiangcheng Cao, Ziqi Cai +3 more
Abstract · 18 pages, 6 figures

A live esports broadcast needs a director who chooses which camera goes on air and a caster who talks over that choice, every second and without knowing what comes next. Streaming video language models have learned to commentate, but only over a broadcast that a human director has already produced. We study live broadcast production, in which a model receives the raw camera of every player and must both direct and commentate from the past alone. We present DirCast, a single streaming video language model that does both. Every second it scans all cameras at low resolution, decides with a hold-or-switch director whether and where to cut, and reads the chosen camera at high resolution before it speaks. Its context keeps what went on air and what was said and lets the other views expire, which bounds the cost of a stream of any length. We build a dataset of 27 professional Dota 2 matches with a camera for every hero, the human director's choices and the casters' commentary. DirCast directs better than rule-based and zero-shot directors including GPT-5, and its commentary wins 29.2 percent of comparisons against GPT-4o under the Proact-VL protocol, against 19.4 percent for the strongest of eleven streaming commentators that watch the human broadcast.

0.33

Class-Grounded Batik Captioning under Scarce Text Annotation: Label Retention and Source-Specific Error Propagation

cs.CV ·2026-10-22 ·Novanto Yudistira, Muhammad Dzaka Aufa Fadhillah, Irawati Nurmala Sari
hot direction
Abstract

Class labels offer compact supervision for describing cultural-heritage images when expert text annotations are scarce. We investigate a batik-captioning cascade that integrates a LoRA-adapted DINOv2 classifier with Qwen2.5-VL-7B, adapted using LoRA and loaded in 4-bit NF4. A $3\times2$ evaluation combines three captioner conditions with two inference prompts to examine adaptation and predicted-class injection. Across 150 caption-evaluation images, all three injected conditions achieve a ground-truth class-label mention rate of 0.9067. The classifier's separate test yields 96.10% pooled accuracy, with 98.33% on 300 public-source images and 75.76% on 33 local images. Supported-class macro-F1 and balanced accuracy on the local subset are 0.8493 and 0.8302, respectively. The classifier and captioner use separate partitions, and their training--evaluation overlap remains unquantified. Differences in label semantics and training support also affect interpretation of the source-specific accuracy gap. Reference-based caption metrics and image--text compatibility scores produce different model rankings. Qualitative examples show both retention of correctly predicted identities and propagation of classification errors into fluent descriptions. These findings characterize how class conditioning preserves label information and transfers upstream errors into batik captions under scarce text annotation.

0.32

SpectralCanary: Embarrassingly Simple Low-Frequency Traces for T2I Dataset-Use Auditing

cs.CV cs.LG ·2026-10-22 ·Sy-Tuyen Ho, Anirudh Satheesh, Soham Chawathe, Huy Nghiem, Yongkyun Lee +2 more
published
Abstract · NeurIPS 2026 Workshop on Attributing Model Behavior at Scale

Auditing whether a text-to-image (T2I) model was fine-tuned on a particular image collection is difficult: a model can absorb collection-level visual characteristics without reproducing training images. Proactive dataset-use auditing plants a shared signal in a dataset before release, then later checks whether that signal appears in a suspect model's generated outputs. Existing approaches often trade off verifiability, fidelity, robustness, and deployability. We introduce $\textbf{SpectralCanary}$, which uses a smooth low-pass mask to blend a fixed canary image into every collection image while leaving captions unchanged and requiring neither auxiliary generation nor per-image optimization. A lightweight verifier detects the transferred signal in suspect-model outputs. Across eight tests on the Pokemon dataset and three realistic data transformations, spanning two architectures families, SpectralCanary achieves the highest average per-generation TPR@$1\%$ FPR ($91.1\%$) and fixed threshold model-level accuracy using 30 generations ($94.8\%$) among the evaluated methods. Across five datasets spanning artistic, medical, satellite, and fashion domains with SD3.5, SpectralCanary averages $92.4\%$ per-generation TPR@$1\%$ FPR while preserving the released images (SSIM $0.989$, LPIPS $0.022$) with a near-zero mean change FID (mean $Δ$FID $0.05$) across datasets. Matched-energy ablations on Pokemon and ROCOv2 support our design choice, low-frequency placement transfers more reliably compared to mid/high-frequency placement. Under the evaluated adaptive attacks, the most effective removal requires aggressive modification and incurs more than $25$ $Δ$FID points. These results identify low-frequency visual signals as effective carriers for collection-level T2I audit traces.

0.32

Lychee: Certified Screening of Inverted Lists for High-Recall Vector Search

cs.IR ·2026-10-22 ·Dongfang Zhao
theory
Abstract · 14 pages, 12 figures

Nearest-neighbor search over embedding vectors matches faces against watchlists, transactions against known fraud patterns, and medical images against confirmed cases, where a missed neighbor is a missed threat, fraud, or diagnosis. Because an approximate index cannot report which queries lost a true neighbor, these applications require recall of 0.999 to 0.9999, at a cost far below that of exact search over billions of vectors. Inverted-file indexes reach high recall by probing more lists, but they read and score every probed vector at full precision, so that throughput falls in inverse proportion to the number of probed lists. Graph indexes saturate below these targets, whereas quantized lists must rerank their candidates with the full vectors. Our insight is to decide for almost every probed vector that it cannot enter the top k of the probed lists without reading the vector, through the certified lower bound of exact search by certified exclusion. This paper turns this insight into Lychee, a GPU inverted-file search that screens the probed lists on tensor cores under a threshold from anchors of the same lists and computes exact distances only for the few survivors. We prove that Lychee returns exactly the k nearest vectors of the probed lists, the answer of an inverted-file search at full precision, while it reads a few hundred full vectors per query regardless of the number of probed lists. On eleven datasets from one million to one billion vectors and against ten GPU and CPU baselines, Lychee is 9.7 to 117 times faster than IVF-Flat at equal recall with 200 probed lists on datasets of up to 100 million vectors. It also reaches recall levels that CAGRA and HNSW do not reach and outruns exact search by certified exclusion by 5 to 18 times above ten million vectors. On four GPUs, Lychee searches one billion vectors at recall 0.9999, 44.5 times faster than IVF-Flat on BIGANN-1B.

0.32

Synthetic-to-Real Code-Switched ASR: A Spontaneous Nepali-English Podcast Benchmark

cs.CL cs.SD ·2026-10-22 ·Firoj Paudel
benchmarkcode/data
Abstract · 9 pages, 4 figures, 7 tables. Code: https://github.com/Firoj-Wise/nepali-english-codeswitch-asr, Dataset: https://huggingface.co/datasets/Sa…

Nepali-English code-switched conversational speech presents severe challenges for automatic speech recognition (ASR), chiefly through script confusion (inappropriate mixing of Devanagari and Latin orthography) and elevated error rates at language switch points. Supervised fine-tuning (SFT) can adapt multilingual foundation models, but human code-switched audio is expensive to collect and annotate. We curate a new benchmark of 13,044 spontaneous podcast utterances (15.45 h total; 10,214 test, 2,830 dev), audited and verified to eliminate machine translation leakages and acoustic loops. We investigate a two-stage synthetic-to-real adaptation framework: Stage 1 applies SFT on 0.95 h of speech synthesized via a multi-speaker, dual-engine TTS pipeline; Stage 2 explores Group Relative Policy Optimization (GRPO) with transcription and script rewards. We find that synthetic SFT provides the primary cross-domain transfer to spontaneous human speech, reducing Mixed Error Rate (MER) from 23.40% to 21.52% (-1.88 pp) and switch-point error rate (SPER) from 35.89% to 28.77% (-7.12 pp). Subsequent GRPO yields small, statistically significant adjustments on 173 test utterances (1.69%), driving English script confusion (SCR_en) from 6.45% to 6.36% (p <= 0.001) and SPER to 28.70% (p = 0.022 clustered). Crucially, ablation reveals that in this SFT-converged encoder-decoder setting, an auxiliary script reward term yields performance identical to MER-only optimization because rollout candidate groups exhibit near-zero script variance on synthetic data. On monolingual benchmarks, SFT improves Nepali MER by 2.71 pp (stabilizing at 18.42% after GRPO, a 2.38 pp gain over baseline) while preserving English lexical WER (3.38% vs. 3.33%). Verified transcripts, timestamps, episode metadata, and reconstruction scripts are publicly released.

0.31

Refining Open-Vocabulary Semantic Segmentation via Regional Semantics and Visual Prototypes

cs.CV ·2026-10-22 ·Aijing Yu, Zhengbo Wang, Lijun Sheng, Jian Liang, Xiao-Yu Zhang
published
Abstract · ICASSP 2026

Contrastive Language-Image Pre-training (CLIP) has shown superior performance in the image-level open-vocabulary classification task, benefiting from its powerful alignment of visual features and text embeddings. However, dense prediction tasks require models to perform pixel-level predictions, necessitating modifications to CLIP for region-text alignment. In this paper, we adjust the attention layers within the vanilla visual encoder of CLIP. With a novel attention calculation method for selective attention features, the model retains more regional semantic information. To further mitigate the negative impact of the modality gap, we introduce additional visual prototypes to refine the original predictor solely based on text embeddings. Specifically, patch features with confident predictions are selected to form an additional auxiliary memory-based classifier for more accurate prediction. We further integrate an optimizable feature extractor with an exponential moving average updating strategy to enhance prediction quality. Extensive experiments on $5$ semantic segmentation datasets demonstrate the effectiveness of our method.

0.31

SyncVid: World-Equivariant Diffusion for Synchronized Video Generation

cs.CV ·2026-10-22 ·Farhad G. Zanjani, Animesh Karnewar, Fatih Porikli, Amirhossein Habibian
code/data
Abstract · 28 pages, 8 figures, 9 tables. Project page: https://qualcomm-ai-research.github.io/SyncVid-Webpage

Generating multiple videos of a dynamic scene from a single reference video requires synthesizing both observed and unobserved content consistently across viewpoints. Existing camera-control methods for video diffusion typically rely on architectural mechanisms, including camera embeddings, cross-view attention, and geometric conditioning. Although effective for static scene content, these approaches often struggle with the fine-grained, dynamic geometry that must remain coherent across views and over time. We synchronize the diffusion process itself by sampling the Gaussian noise prior in a canonical 3D space derived from the source video and rasterizing it into target cameras, rather than sampling independently in each 2D image plane. Our method, SyncVid, lifts a monocular video into a canonical 4D representation, associates noise with scene geometry and unseen space, and rasterizes the resulting structured noise into arbitrary views. Because rendering can introduce local correlations, we further propose a cross-view-consistent whitening procedure that decorrelates neighboring pixels while preserving correspondence across views. The structured noise serves as the diffusion prior for a LoRA-adapted pretrained video diffusion model, requiring no architectural modifications. By enforcing spatiotemporal and cross-view coherence at the source of stochasticity, SyncVid generates synchronized videos and improves both perceptual quality and cross-view consistency in our experiments. Project page: https://qualcomm-ai-research.github.io/SyncVid-Webpage.

0.31

Memory-Aware Reward Modeling: Learning Human Preferences for LLM Memory Use

cs.CL cs.AI ·2026-10-22 ·Yijiang River Dong, Zexue He, Cheng Yu, Tiancheng Hu, Michael J Ryan +7 more
Abstract

Memory enables language model assistants to maintain continuity across sessions and generate personalized responses, but knowing more about a user does not necessarily produce responses they prefer. Existing evaluations assess memory capabilities without directly establishing whether memory use benefits the user. To study this gap, we collect approximately 1,000 pairwise judgments from 87 participants using their own ChatGPT queries and personal memories. We find that memory-conditioned responses are not consistently preferred over responses generated without memory and in some cases, users strongly prefer the latter. Automatic evaluators also show limited agreement with users and favor memory-conditioned responses substantially more often than users do. Motivated by these gaps, we introduce a memory-aware reward model that learns from user judgments, achieving 72.8\% preference prediction accuracy and outperforming the existing baselines. Our findings motivate evaluation and optimization objectives that align memory use with the preferences of the people whose information is being used.

0.31

Making MLLM Latents Matter: Factor-Aware Predictive Learning for Cinematic Video Generation

cs.CV ·2026-10-22 ·Songlin Yang, Zhe Wang, Xiaotong Zhao, Xi Wang, Ran Zhang +6 more
community pick
Abstract

Cinematic video generation requires both temporal inference from limited inputs (i.e., sparse rather than frame-by-frame guidance) and factor-aware control of character, scene, action, and camera. Multimodal large language models (MLLMs) offer useful priors for both capabilities, but their representations are not automatically actionable for video generation: a Video Diffusion Transformer (DiT) may bypass them through native conditioning pathways, and using them does not ensure either factor-aware control or generation-compatible temporal prediction. We propose Factor-Aware Predictive Learning (FAPL), which establishes effective MLLM latent use as a shared prerequisite and then targets the two capabilities with complementary learning signals. First, latent-only generator grounding trains the DiT with a frozen MLLM and connector, strengthening the DiT's use of MLLM latents for generation. Second, for factor-aware control, paired videos specify a target-factor change and non-target preservation, and factor-span replacement trains the DiT to realize this relationship in its output. Third, for temporal inference, text-only and text-plus-complete-video observations share a video target, and the trained DiT is then frozen while the MLLM and connector are optimized through it. This final stage trains limited-input representations for generation while retaining complete-observation reconstruction as a complementary training mode. Extensive experiments demonstrate our consistently better performance on three cinematic video generation tasks (i.e., sparse-to-video, factor-aware editing, and multi-shot generation) over baselines and validate FAPL's contributions to MLLM feature use, temporal control, and factor coordination.

0.30

QICOM: Open-Source Software for Reproducible Image Quality Evaluation, Blinded Assessment, and Radiomic Comparison

eess.IV cs.AI ·2026-10-22 ·Fabio Mattiussi, Switinder Ghotra
code/data
Abstract

QICOM (Quality Image COMparator) is open-source software for quantitative image comparison, paired radiomics, and blinded assessment. It saves processing settings, image identities, measurements, and response mappings so a comparison can be reopened and recalculated. We evaluate reference agreement, responses to controlled perturbations, and recovery of saved analyses. Synthetic experiments cover 37 quantitative descriptors using independentnumerical references and separate upstream integration comparisons. The QICOM/MIRP backend meets the reference tolerances for all 482 numeric IBSI 1 digital-phantom entries and all 1,622 entries across configurations A-E of the public CT reference case; these comprise 1,804 feature/aggregation comparisons and 300 preprocessing diagnostics. Restricting masked SSIM and HFEN to complete kernel support removes the observed dependence on excluded pixels. In a synthetic filtering example, stronger smoothing increases SSIM while worsening MSE and HFEN relative to milder smoothing and changing GLCM contrast more strongly. Reopened radiomics projects reproduce all 43 feature pairs per comparison, and two blinded deliveries return four correctly mapped scripted responses. These experiments document the calculations and data transfers linking the three workflows.

0.30

GraphCAD: Geometry-Aware Diffusion on Edge-Adjacency Graphs for CAD Generation

cs.CV ·2026-10-22 ·Wenhao Zheng, Xianggen Liu, Li Yue, Jiancheng Shi, Jiancheng Lv
top venuedataset
Abstract

Generative modeling for Computer-Aided Design (CAD) remains fundamentally challenged by the need to capture the intricate interplay between discrete topological connectivity and precise spatial geometry inherent in Boundary Representations (B-reps). Existing generative approaches largely formulate CAD construction as a sequence modeling problem, but consequently lack an explicit topological inductive bias. In this work, we present GraphCAD, a geometry-aware generative framework that centers modeling on the Edge-Adjacency Graph (EAG), a novel graph-based CAD representation tailored for generative tasks. GraphCAD adopts a hybrid graph diffusion formulation to jointly model the discrete and continuous components of EAG and introduces a geometry-aware denoising mechanism that renders node attributes into spatial point sets. These point sets are encoded by an SE(3)-equivariant backbone and fused into the node representations, thereby grounding the generative process in physical geometry. To support this study, we construct EAGBench, a large-scale dataset of engineering CAD models. Extensive experiments show that GraphCAD can generate complex CAD models with more than 50 faces and achieves state-of-the-art performance in geometric fidelity and distributional alignment compared with sequence-based baselines.

0.30

A Visual Grounding Benchmark for Retinal OCT with Multimodal Large Language Models

cs.CV ·2026-10-22 ·Ronald Fecso, José Morano, Botond Fazekas, Paul Widmann-Sedlnitzky, Dorota Himel +4 more
benchmarkcode/data
Abstract · Under review at npj digital medicine

Multimodal large language models (MLLMs) can answer clinical questions about medical images, but trustworthy use requires visual grounding: models should localize the pathology they report rather than rely on spurious shortcuts. Whether MLLMs can visually ground clinical findings in retinal optical coherence tomography (OCT), the most widely used imaging modality in ophthalmology, remains largely unexplored. To address this gap, we introduce RetinaGroundBench, the first benchmark for biomarker visual grounding in retinal OCT. It comprises 815 B-scans exhaustively annotated by ophthalmologists with 4,537 bounding boxes covering 13 biomarkers across 8 diseases. Because MLLMs often fragment a single lesion into multiple partial predictions, we propose AggIoU, a metric that evaluates whether the union of overlapping predicted boxes covers the annotated region. Across 27 open-source MLLMs (0.8-38B parameters), we find that visual grounding remains largely unsolved: the best general-purpose model achieves a macro F1 of just 23.7%, with consistently low recall, despite reaching up to 73% balanced accuracy on biomarker presence questions. To narrow this gap, we developed RetGrounder, an MLLM fine-tuned using an OCT-specific strategy integrating grounding and textual information. Despite its relatively small size (8B), RetGrounder improves grounding F1 to 33.9%, a 43% relative improvement. Our benchmark provides a reference point for measuring progress in the field of OCT diagnosis with MLLMs.

0.30

AgentDreamer: Agentic 3D Motion Reasoning for Text-Controllable Physics Simulation

cs.CV ·2026-10-22 ·Mijeong Kim, Suho Park, Junha Park, Bohyung Han
code/data
Abstract · Project page: https://cv.snu.ac.kr/research/AgentDreamer

We propose AgentDreamer, an agent-guided optimization framework that aligns text instructions with simulated motion through explicit 3D motion planning. Given a text instruction and scene geometry, an off-the-shelf agent translates the instruction into a structured semantic motion prior, specifying what should move, in which direction, and in what temporal order. The agent represents this prior using parametric functions over a series of temporal chunks, yielding dense 3D trajectories. We then optimize the parameters of a differentiable MPM simulator such that its rollout best matches the prior while remaining constrained by the scene dynamics. This allows the agent to provide semantic guidance from language, while the simulator grounds it in physically feasible motion. We further introduce Motion-IG, an information-theoretic metric for text-motion alignment based on masked text recovery. By comparing the likelihood of recovering masked motion-related content with and without the simulated motion, Motion-IG measures the information contributed by the motion beyond that available from the remaining text and scene context. Experiments show that AgentDreamer improves text alignment over video-based guidance, producing simulations that more closely follow the requested physical behavior.

0.30

What the Specification Did Not Say: Specification-Oracle Gaps in Test-Graded Coding-Agent Benchmarks

cs.CL ·2026-10-22 ·Fei Ding
benchmark
Abstract · 13 pages

Coding-agent benchmarks often describe a task in natural language but grade it with tests taken from the upstream project. Passing those tests can require information the description never states: a constant's name, a module path, an exact message. We call this a specification-oracle gap: where it exists, a benchmark also measures whether an agent reproduces upstream implementation details. We formalise the gap, including the inconsistencies that chains of dependent tasks add, and build a static audit for five languages that classifies every name-, path- and literal-level dependency of every scored test (held-out precision 89.2%, recall 92.9%). On SWE-Milestone (98 graded milestones, 35 published runs), 25 milestones contain a test that depends on an unstated, non-inferable name or path. Build units with such a dependency fail in 80.6% of runs, against 48.4% for comparable units without one, and withholding a single hidden name from the reference stops its unit's tests from compiling in 126 of 128 cases. The leaderboard's ranking is robust to removing the affected tests, but they shift scores within repositories and favour agents that reproduce upstream names. Across four further benchmarks, prevalence appears to follow construction: a third or more of the tasks are affected where specifications are release notes or LLM summaries, almost none where tasks are screened or written for their tests; how much a gap costs depends on how the harness grades. We release the audit, corrected filter lists and specification addenda.

0.30

Once a Tree, Always a Tree? Cross-lingual Transfer of Hierarchical Generalization in Language Models

cs.CL ·2026-10-22 ·Shin Yamaguchi, Masaya Taniguchi, Tatsuki Kuribayashi, Taro Watanabe, Yoshinobu Kano
top venue
Abstract · EMNLP 2026 main

Language models (LMs) often exhibit successful cross-lingual transfer through continued (pre)training, still their underlying process remains unclear. One possibility is that LMs can leverage the evidence toward language-universal principles, namely, hierarchical structure, from a source language (L1) to a new language (L2), while many other non-linguistic explanations, e.g., improved initialization of position embeddings, can be potentially possible. In this study, to quantify the cross-linguistic transferability of linguistic evidence in cross-lingual LM training, we introduce a new experimental design, namely, a cross-lingual poverty-of-stimulus (CPOS) setting that ablates how LMs' hierarchical generalization in L2 can be led by structural evidence provided in L1. We contrast counterfactual L1 and L2 learning scenarios where the L2 corpus is ambiguous between linear and hierarchical grammar rules, while the amount of positive evidence for the hierarchical rule is controlled on the L1 side. Our experimental results show that LMs can partially, yet not perfectly, transfer L1 structural evidence to L2, yielding only a small but detectable hierarchical bias, while the strength of transfer depends on language pairs. We further ablate the conditions for successful transfer, motivated by recent reports on the substantial impact of vocabulary overlap. We indeed find that vocabulary overlap is crucial even for the cross-lingual transfer of hierarchical syntactic generalization. This suggests that this also explains the language-pair-dependent tendencies in the main experiments and suggests that the transfer of linguistic evidence in continued new-language learning is only conditionally achievable.

0.30

Where meanings converge and diverge: Revisiting linguistic relativity across 81 languages

cs.CL ·2026-10-22 ·Weiran Wang, Wenjuan Qin, Haitao Liu
unusual
Abstract

Do people who speak different languages organize concepts in similar ways? This question lies at the heart of the Sapir--Whorf hypothesis and debates about linguistic relativity. Shared basic meanings, however, need not imply shared relations among them. We compared relations among 196 basic concepts across 81 languages, using word embeddings from two multilingual models to quantify conceptual structure expressed in language. Cross-language correspondence was greater for sociocultural meanings than for meanings concerning nature and motion, a contrast recurring across models and classifications. Within the sociocultural domain, relational differences increased with geographic separation; less closely related languages differed more across all domains. These findings suggest that universality and relativity are interwoven across conceptual domains. They inform an account in which shared cognitive resources interact with linguistic and local experience, and predict greater cross-language correspondence in human semantic judgments for sociocultural than for nature and motion meanings.

0.29

A Grammar of French Civil Cassation: A Ground-Level Dataset and Tool

cs.CL cs.AIcs.CY ·2026-10-22 ·Avrile Floro, Tamara Dhorasoo, Nils Holzenberger, Thomas Le Goff, Tiphaine Viard +1 more
publisheddataset
Abstract · Accepted at JURIX 2026 (39th International Conference on Legal Knowledge and Information Systems)

This paper presents a dataset of 121,536 ground-reasons pairs across 86,464 rulings of the five civil chambers of the French Court of Cassation (2016-2025). An appeal on points of law cannot itself be appealed and often decides a case for good. Understanding how the Court rules is thus crucial and can help lawyers who draft those appeals. Inside a ruling, every ground of appeal (a distinct legal complaint against the appealed decision) has its own answer from the Court. This answer is called the reasons and it belongs to one of a few doctrinal families. Using a deterministic pipeline, we pair each ground with its reasons and attach the pair to its family, according to the established French taxonomy. We obtain a descriptive grammar of civil cassation that shows which grounds succeed, which legal texts they refer to, and how both evolve over the decade. We release the dataset, the code and an interactive consultation tool.

0.29

MICE to Meet You: From Re-ranking Only to End-to-End Retrieval

cs.IR ·2026-10-22 ·Victor Morand, Yuxuan Zong, Mathias Vast, Josiane Mothe, Benjamin Piwowarski
code/data
Abstract · code: https://github.com/xpmir/mice

Late-interaction models offer state-of-the-art efficiency-effectiveness trade-offs for neural retrieval by enabling document representations to be pre-computed and efficiently indexed. Yet, their relatively simple interaction mechanisms, typically MaxSim, can limit their effectiveness compared to cross-encoders. MICE, a recently introduced re-ranking architecture, aims to combine cross-encoder-like effectiveness through expressive transformer-based scoring with reduced inference-time computation through document-side pre-computation. However, no practical way to store these document representations was explored. We address this limitation here and introduce MICE-P, the first end-to-end retrieval pipeline with MICE, based on PLAID to index the contextualized document token representations. We conduct a controlled comparison of MICE, ColBERT and cross-encoders by training comparable models using the Ettin Family under matched experimental conditions. Our results show that MICE-P defines a new class of retrieval architecture with an efficiency-effectiveness trade-off of its own: it matches late-interaction retrieval throughput while retaining the effectiveness of cross-encoders.

0.29

MixPT: Point Transformer with Interleaved Window and Global Attention

cs.CV ·2026-10-22 ·Minkyun Seo, Hyungtae Lim, Jaesik Park
code/data
Abstract · 23 pages, 14 figures, 12 tables. Project page: https://minkyunseo.github.io/MixPT/

Point cloud transformers aim to capture detailed local geometry and relationships across a scene at a manageable computational cost. Two common approaches are window attention and global attention. They differ in how they enable scene-wide interaction, which refers to information exchange among tokens across a scene. Window attention reduces computational cost by restricting self-attention to groups of tokens called windows; in serialized designs, each window contains a fixed number of consecutive tokens from a spatially ordered sequence. Because each block only directly connects tokens within the same window, information exchange across windows relies on changes in window membership across blocks and hierarchical pooling. Global attention, in contrast, directly connects all tokens within a scene, but computing attention over all token pairs can be costly when the number of tokens is large. To balance computational cost and scene-wide interaction, we introduce MixPT, a point transformer that uses a mix of window and global attention blocks in an interleaved schedule, while retaining a single fixed serialization pattern for its window blocks. For scenes with a large number of tokens, MixPT applies global attention to spatially pooled tokens and returns residual updates to the original-resolution stream. Across indoor and outdoor benchmarks, MixPT achieves competitive accuracy with lower latency than representative window and global attention baselines. We further demonstrate the feasibility of scaling the attention stage in depth by repeating the interleaved pattern of window and global attention blocks.

0.29

SAGE: Sense-Aligned Multimodal Grounding Evaluation with Gloss-Augmented Reasoning

cs.CL cs.CV ·2026-10-22 ·Sunjae Kwon, Suhaas Kiran DG, Zonghai Yao, Talha Chafekar, Deepali Singh +1 more
benchmark
Abstract

This paper introduces Sense-Aligned Grounding Evaluation (SAGE), a unified 10-candidate image-selection benchmark that reformulates VWSD, VerSe, and WikiDiverse into a shared term-centric protocol covering noun-, verb-, and entity-centered grounding. SAGE includes confusable hard negatives and a context-swapped VerSe diagnostic to probe target-term grounding beyond direct caption--image compatibility. We further propose Gloss-Augmented Reasoning (GAR), a training-free framework combining definition grounding with constraint-guided semantic verification. Experiments show that strong VWSD performance does not transfer uniformly across grounding settings. With Gemma3-12B, definition augmentation improves all three subsets, while constraint-guided reasoning improves VWSD and yields particularly strong gains on VerSe, but is less effective on WikiDiverse. Combining the two signals achieves the best macro-average performance, improving Hits@1 from 65.90 to 71.22. These results show that SAGE supports both unified evaluation and fine-grained analysis of explicit semantic reasoning across grounding settings.

0.29

Cola2: Co-Adaptive Hierarchical Language Modeling with Continuous Semantic Compression and Balanced Discrete Generation

cs.CL ·2026-10-22 ·Hongcan Guo, Yian Zhao, Feng Wang, Shen Nie, Guoqiang Wei +2 more
Abstract · 47 pages, 14 figures, 5 tables

Hierarchical language generation separates continuous semantic modeling from discrete text realization, but its effectiveness depends on what the latent space compresses and how the discrete generator uses it. We propose Cola2, a framework that jointly studies continuous semantic compression, the balance between reconstruction and generation, and continuous-discrete co-adaptation. Cola2 dynamically segments text into semantic spans, controls the compression rate through auxiliary-loss-free negative feedback, and combines reconstruction and generation supervision to learn compact, predictive latent representations. A block-causal masked diffusion decoder is trained with clean, noisy, and dropped latent conditions, while classifier-free guidance adjusts the balance between semantic reconstruction and autonomous generation. Token-prediction gradients update the encoder through the latents, whereas Flow Matching prior learning is decoupled from representation learning by stop-gradient. During inference, generated text is re-encoded under a fixed span layout to keep the latent history consistent with the actual text context. Experiments on five downstream tasks with about 210 billion training tokens validate the core designs and show competitive, unsaturated scaling. With 346.03 billion source tokens, an unconverged checkpoint reaches a 60.10 average across eight tasks and surpasses Llama-3-8B on HellaSwag and StoryCloze despite its reported pretraining budget being over 43 times larger. These results demonstrate the potential of continuous-latent hierarchical language generation.

0.29

Scaling cross-situational word learning on children's naturalistic experiences

cs.CL ·2026-10-22 ·Michael C. Frank, Alvin W. M. Tan, Khai Loong Aw, Robert Z. Sparks, Victor Moreno +7 more
Abstract

How do children learn the meanings of the simple, concrete words that form a majority of their early vocabulary? Statistical information about many word meanings can be recovered from the gradual accumulation of information about the co-occurrence of words with their referents, a proposal known as ``cross-situational learning''. Contrastive vision--language models allow this learning strategy to be tested on egocentric video data capturing children's early real-life experiences with visual and linguistic inputs. Initial successes of this method on small datasets provide a proof-of-concept, but vision models struggle to learn robust visual representations when trained on these same egocentric videos. How dependent is cross-situational word learning performance on the quality of the model's visual representations? And can cross-situational models scale to larger and more diverse datasets? To address these questions, we trained models on the BabyView corpus, including 2,441 hours of longitudinal egocentric video from 48 children. Models trained on this dataset learned object--label mappings, but performance was far higher when they used internet-pretrained visual representations as opposed to when visual representations were learned solely from the developmental video data. Further, the majority of the learning signal was carried by a small proportion of utterances that described a visible referent. Models learned lexical representations that showed recognizable human-like semantic structure, but model performance showed substantially different trajectories over training experiences than those observed in children. Cross-situational learning is a viable word learning strategy for early noun vocabulary, but constructing developmentally plausible models will require understanding how to learn high-quality visual representations from naturalistic data.

0.28

Automated Enrichment Analysis with GraphRAG: Efficient Multi-Database ORA Extraction and Biological Interpretation

cs.IR cs.DB ·2026-10-22 ·Guangcan Li, Shuyuan Zheng, Jiayao Liu, Shigeto Seno, Makoto Onizuka
published
Abstract · Accepted to the 2026 IEEE International Conference on Bioinformatics and Biomedicine (IEEE BIBM 2026)

Over-representation analysis (ORA) is widely used to interpret gene-expression studies, but its outputs are often fragmented across gene-set databases and presented as flat term lists that obscure gene-level links among related enrichment signals. We propose Graph-Based Over-Representation Retrieval-Augmented Generation (GBOR-RAG), an integrated and traceable framework for ORA interpretation. GBOR-RAG represents heterogeneous gene-set databases as a unified term--gene knowledge graph, where shared genes connect related terms across sources. After graph-based ORA scoring, the framework retrieves a compact connected evidence subgraph that preserves both cross-database relationships and gene-level support. The subgraph conditions an LLM to generate biological interpretations, and we introduce a claim-level faithfulness metric to assess whether generated claims are supported by explicit evidence. Experiments show that graph-based ORA preserves conventional ORA results with high consistency (mean Pearson $r=0.997$) while reducing query time by a median of $37\times$ (up to $48\times$). The retrieved subgraphs average 36.8 nodes and achieve the highest overall integration score (4.12), and GBOR-RAG attains the highest claim-level faithfulness (0.90) among the compared systems. Adding a PubMed literature layer further improves the overall interpretation quality from 4.17 to 4.75.

0.28

TRIAD: Benchmarking Omni-Modal Ambiguity in Multimodal Large Language Models

cs.CL ·2026-10-22 ·Zhaolu Kang, Yidi Wang, Yile Li, Siqi Zeng, Yingjie He +2 more
publishedbenchmark
Abstract · NeurIPS 2026 Evaluations and Datasets Track

Ambiguity is a central feature of natural communication: a sentence, image, or sound may remain underspecified until evidence from the other channels is considered. Yet current multimodal benchmarks mostly test perception, recognition, or reasoning over already determinate inputs, leaving unclear whether omni-modal large language models can resolve ambiguity distributed across text, vision, and audio. We introduce TRIAD, a benchmark for omni-modal ambiguity resolution. Each item is a text--image--audio triplet paired with a question and a set of answer options, and is constructed so that the full triplet determines a unique answer while removing any one modality makes the set of consistent options non-unique. TRIAD contains $327$ hand-written queries in $86$ scene groups and covers $18$ ambiguity categories across the three modalities. Evaluating $14$ omni-modal systems reveals that strong item-level performance does not translate into robust scene-level disambiguation: the best model still trails humans by over $20$ points in group accuracy. Removing answer options widens the gap further, driving model group accuracy to the single digits while humans remain robust. Leave-one-out and transcript/description counterfactuals further show that current systems often underuse audio cues when text and image are present. TRIAD reframes omni-modal evaluation around disambiguation rather than recognition, and provides a taxonomy, benchmark, and diagnostic protocol for measuring this capability.

0.28

Compiling Autoregressive LLM Control into Policy-Exact Parallel Inference for Hierarchical RAG

cs.CL cs.IR ·2026-10-22 ·Zehao Lin, Qianfeng Shen, Xixuan Hao, Shaobo Cui, Chunyu Li +6 more
Abstract · 5 pages, 2 figures, 1 table

Large language models increasingly act as controllers in structured decision pipelines, yet their decisions are still executed by autoregressive text decoding. Hierarchical RAG is one such setting: at each frontier of a document tree, an LLM commits to $b$ of $k$ sections under an input-dependent action grammar, executed as either failure-prone free-form JSON or reliable but strictly serial grammar-constrained generation. We show that this serialization is unnecessary for bounded-horizon control: the reachable future forms a shallow action-prefix tree, which we compile --- one root prefill plus one tree-causal probe evaluate every prefix-conditioned action distribution with serial decoding's tokens, positions, and causal ancestry. A prefix-state lemma makes compiled control policy-exact in exact arithmetic: the same policy with a two-forward critical path and zero output tokens. One further probe pass certifies the exact order-marginalized MAP set. Across four document-disjoint splits and two scales on the primary MetaX device, compiled control matches the serial policy on $84$--$89\%$ of questions --- every instrumented divergence a floating-point near-tie --- and cuts control latency by $35$--$47\%$ versus token-wise serial decoding and $6$--$7\%$ versus an idealized chunked engine; an H800 repetition preserves the ordering.

0.28

Mostly Reading: What Open-Book Scores Measure in Document-Level LoRA Adapters

cs.CL cs.AIcs.LG ·2026-10-22 ·Martino M. L. Pulici, Cuong Xuan Chu, Evgeny Kharlamov, Volker Tresp
Abstract · 27 pages, 3 figures, 15 tables, under review at ACL ARR; v2: minor abstract fix

Single-document LoRA adapters have been proposed as units of knowledge to be merged and retrieved; yet open-book scoring, with the document in the prompt, can credit a document's own adapter (its specialist) for reading that document as readily as for holding it. Training one adapter for each of 150 Wikipedia articles on Llama-3.1-8B-Instruct, we find that open-book scores mostly measure a reading behaviour every adapter shares: a random adapter trained on another document recovers 82 % of the specialist's open-book advantage over the base model but only 2 % of its advantage on a cloze probe that asks for the same document's facts without providing the source text, a share we cannot distinguish from zero. Content does reach the weights, since the specialists beat the base model in cloze recall by 0.094, and the geometry of their updates follows the documents they read; yet, in a test of 30 queries, adapters chosen by weight-space proximity or text embedding are not distinguishable from random ones on either probe, although that test cannot exclude an advantage of about four fifths of what a query's own adapter gains over a random one. At 16 documents or more, correctly scaled merges recover 87 % to 89 % of the specialists' open-book advantage but only 12 % to 16 % of their cloze advantage, with intervals that run from below zero to at most 32 %, against 94 % for joint training. We also report four silent errors, each of which produced a clean, wrong result.

0.28

OCRia-1 Technical Report

cs.CV ·2026-10-22 ·Aman Rangapur, Antonio Jimeno Yepes, Daniel Bustamante, Nutchapol Dendumrongsup, Sebastián G. Acosta +2 more
hot direction
Abstract

We introduce OCRia-1, a 4B vision-language model (VLM) for document parsing that excels at preserving table structure, extracting chart data and retaining text formatting. OCRia-1 handles content that has traditionally been challenging in document parsing: merged cells and spanning headers, tables with thousands of cells, chart data, and meaningful text formatting such as bold, italics, subscripts and superscripts, preserving details needed for business-critical document analysis. At the time of this report, it achieves the highest scores on tables, charts and text formatting among the systems evaluated, including a 35B model and Mistral OCR 4.1. Its mean ParseBench score over the four dimensions we evaluate is 0.87, against 0.75 for the next-best system. We emphasize data quality and reward design in training OCRia-1. Synthetic pages provide exact ground truth for long tables and rich formatting, while real pages are transcribed using a pipeline grounded in the PDF text layer. We attribute OCRia-1's performance primarily to careful data curation that prioritizes quality and diversity over quantity. Dimension-specific rewards evaluate table reconstruction, chart-data extraction and text-formatting fidelity. We train the model on the curated data using supervised fine-tuning (SFT), then refine it through reinforcement learning with verifiable rewards (RLVR). To our knowledge, OCRia-1 is the fastest published document-parsing VLM in its class, decoding at about 4,700 tokens per second with a single request on one B200 GPU, about nine times faster than the base model on stock vLLM. This is achieved through a block-diffusion speculator, quantization and custom kernels.

0.28

IFC-SSL-Bench: Benchmarking Self-Supervised Learning for Annotation-Efficient Multispectral Imaging Flow Cytometry

cs.CV ·2026-10-22 ·Angelo Jovin Yamachui Sitcheu, Roxana-Teodora Fetescu, Thomas Lautenschlager, Nils Friederich, Linus Schilpp +4 more
benchmark
Abstract · 35 pages, six main figures, seven supplementary figures, five tables and four supplemntary tables

Imaging Flow Cytometry (IFC) generates large multispectral datasets, yet the need for expert annotations bottlenecks the development of supervised deep learning models. While Self-Supervised Learning (SSL) offers a solution, its application to IFC remains fragmented and lacks systematic benchmarking. To address this, we present IFC-SSL-Bench, a modular framework for evaluating SSL in IFC. We benchmark six SSL approaches across four backbones on two heterogeneous datasets. By assessing downstream classification performance across annotation regimes of the training sets (1%-100%), we quantify annotation efficiency and identify optimal configurations for high-dimensional biological imaging. SSL matches the performance of fully supervised baselines trained on 100% of annotated data while using as little as 5% of annotations for morphological classification, with advantages in low-data regimes below 5%. Across both datasets, the contrastive approach MoCo-v2 paired with ConvNeXt backbones emerged as the top-performing configuration, with ConvNeXtV1-T best for immunological synapse classification and ConvNeXtV2-T for morphological classification. We identify a clear pattern in transfer dynamics: ImageNet initialization benefits morphological classification, consistent with its visual similarity to natural images, whereas random initialization better captures the more domain-specific immunological synapse classification. SSL elevates class-specific F1-scores by over 35 percentage points for structurally complex classes at 1% of annotated data. We synthesize these insights into a unified set of guidelines for deploying self-supervised workflows in high-throughput IFC, and release the benchmark as a resource for future work.

0.27

Invisibilization as a Source of Probability Bias in Language Models

cs.CL cs.CY ·2026-10-22 ·Christian Arnold, Ernesto Calvo, Guadalupe Gonzalez, Sebastián Vallejo Vera, Tiago Ventura
Abstract · 17 pages, 4 figures, 2 tables

Language models are aligned to avoid using protected attributes such as race, ethnicity, and gender to inform their responses. Therefore, they increasingly refuse or hedge when such attributes appear in a prompt. However, if a system must decide how much to learn from what people tell it, it is costly to refuse valuable information because of its source. In this article, we measure the cost of declining information in a two-urn game with an exact Bayesian answer. We present leading LLM models from two developers (OpenAI and Anthropic) with a prompt that requires them to decide how far to trust an unverifiable report from a participant whose description we randomize across ten configurations (6,600 sessions). We show that leading models often decline to assess the reporter, treating information as worthless at rates set by the description rather than by the evidence. At its widest, models decline a Black-described reporter 46% of the time and a white-described reporter 9%; yet, when they do assess the information, the Black-described reporter receives slightly more trust. The bias is invisible to audits that compare average trust or look only at assessed cases. It grows with reasoning effort, follows words rather than categories (models invoke race for "Black" but not for its Spanish translation negro, and never for "Indigenous"), and worsens under warning instructions a practitioner would try first. We also present a scoring option that reduces the refusal rate bias. When a model's trust is scored against revealed truth over repeated rounds, it credits identical records equally. In four of five configurations from Anthropic and OpenAI, the racial gap in trust falls below 0.02 by the last rounds.

0.27

Measuring the Verifier's Floor: How Much Apparent Rare-Word Text-to-Speech Error Is ASR Error?

cs.SD cs.CL ·2026-10-22 ·Kirill Borodin, Vasilii Kudryavtsev, Nikita Vasiliev, Grach Mkrtchian
Abstract · Submitted to IEEE ICASSP 2027

Text-to-speech (TTS) is usually evaluated by transcribing its output with an automatic speech recognition (ASR) system, so the word error rate also counts recognizer errors. We measure these errors on human recordings of the test sentences. On LibriSpeech and FLEURS, the evaluation recognizer misses 33-67% of very rare words even in human speech, and all three recognizers miss 22-42%. Checking every word with two more recognizer families lowers the apparent TTS error on words that at least one recognizer hears two- to three-fold (very rare LibriSpeech words: from 10-13% to 3-5%), yet very rare words remain the hardest. Ignoring words the recognizer misses in human speech is biased: the other recognizers find TTS errors far more often on these words (4-6% vs. 0.2-0.3%) and overturn about half of the errors this mask keeps. As half of the remaining adjudicated errors may be spelling variants, a listening test is needed.

0.27

Silicon Samples Match the Answers but Not the Respondents

cs.CL cs.CY ·2026-10-22 ·Yusuke Kumagae, Masahiro Kaneko
survey
Abstract

Silicon sampling generates survey respondents by having a large language model (LLM) answer a survey as a person with given demographic attributes, and the resulting synthetic populations are increasingly used in place of human ones. Existing evaluations focus on individual-answer accuracy, per-question answer distributions, and correlations across questions. None of these compares the respondents themselves, so a generator whose respondents have combinations of answers found in no human respondent can still score well if it reproduces every answer distribution. We propose the respondent coverage gap (RCG), which represents each respondent as a point defined by their complete answer profile and measures how far each human respondent is from the nearest generated one, relative to the distance between human samples. On two US social surveys, all six LLMs we evaluate leave human respondents uncovered, and under our main settings, they fall behind a baseline that samples each answer independently from the human marginals. Across the prompting and inference settings we test, every LLM has a higher RCG than a Gaussian copula fitted to the human data.

0.27

GWDM: A Guided Wavelet Diffusion Model for Realistic Histopathology Image Generation

eess.IV ·2026-10-22 ·Abdullah, Tao Huang, Ickjai Lee, Lei Bi, Euijoon Ahn
benchmark
Abstract · 14 pages, 5 Figures, 5 Tables

Histopathology images are recognized as a reliable and comprehensive modality for Deep Learning (DL) based diagnosis. However, effectiveness is constrained by limited large-scale annotated datasets, as collecting expert-labeled medical data is costly and time-intensive. Recent advances in generative models, particularly diffusion models, demonstrate strong capabilities in synthesizing medically relevant content for dataset augmentation. Existing generative frameworks treat histopathology patches as generic natural images, failing to enforce biological constraints such as histomorphological structures and heterogeneous staining patterns essential for valid synthesis. We propose a guided wavelet diffusion model (GWDM) framework that captures image details at multiple scales using wavelets and leverages controllable conditioning based on domain-specific histopathological features (e.g., tissue patterns and color variations). We further formulate a histomorphological guidance loss applied to real and generated patches. A loss ramping strategy gradually introduces these constraints during training, ensuring stable optimization and robust modeling of tissue architecture. GWDM was evaluated on three benchmark datasets (BCSS, MoNuSeg, and PAIP2019), consistently outperforming state-of-the-art generative methods. Ablation studies confirmed that domain-guided conditioning, perceptual loss, and structural similarity loss each contribute meaningfully to synthesis quality. Downstream nuclei and tissue segmentation experiments further validated the clinical utility of GWDM-generated images, demonstrating consistent performance improvements when synthetic data augments real training sets. This work shows that domain-specific conditioning and dual-loss optimization generate realistic histopathology patches, reducing data scarcity and improving clinical downstream segmentation performance.

0.27

Poolwise: Evaluating Reward Models for Selection across Prompts

cs.CL cs.AI ·2026-10-22 ·Mateusz Modrzejewski, Tomasz Radzikowski
Abstract · 38 pages, 9 figures, 47 tables. v2: reorganized around Poolwise, a protocol for evaluating reward models on pooled selection across prompts;…

Reward models are used to select training data, but common benchmarks compare responses only within the same prompt. Selecting data across prompts also depends on how scores compare between prompts. We introduce Poolwise, a protocol that selects a fixed fraction of a response pool by score, measures the selected set's quality using external labels, and reports within-prompt ranking separately. In matched experiments on two rating sets and two model sizes, training each prompt's mean score to match its average response rating improves selection over permuted or omitted mean targets in every seed at 30 percent retention. In grouped HelpSteer experiments, the model with the highest within-prompt accuracy selects a lower-rated set than the correct-mean model. Subtracting prompt means estimated from separate reference responses preserves local rankings but lowers selected helpfulness for six open reward models and two closed judges on HelpSteer. Poolwise provides a direct way to compare reward models and score transformations for data selection.

0.27

A Systematic Comparison of Search Space Complexity and Efficiency for Time-Synchronous End-to-End ASR Models

cs.CL ·2026-10-22 ·Simon Berger, Larissa Kleppel, Ralf Schlüter
published
Abstract · Accepted at IEEE SLT 2026. 7 pages, 4 figures

Search is a central component of Automatic Speech Recognition (ASR) systems and has been deeply investigated for classical ASR modeling based on Hidden Markov Models (HMMs). However, in work on current sequence-to-sequence ASR architectures it is often simplified to a small fixed-size beam search without external Language Model (LM). In this work, we present a comparative analysis of search space complexity and efficiency for Connectionist Temporal Classification (CTC) and neural transducer models under different pruning strategies and LM configurations. We show that these end-to-end (E2E) models work with a small search space most of the time but exhibit large temporary peaks in hypothesis count during periods of uncertainty. Score-based pruning dynamically adapts the search to model uncertainty, leading to a significantly lower Real-Time Factor (RTF). Our results highlight the limitations of search with low fixed-size beam pruning and provide practical guidelines for efficient search in E2E ASR.

0.27

LAMB: A Modern Contextual Language Model for Latin

cs.CL ·2026-10-22 ·Aidan McCarthy
benchmark
Abstract

This article introduces Latin Modern BERT (LAMB), an encoder for the Latin language based on the ModernBERT architecture, pretrained in two phases on more than 74B Latin tokens using a novel morphologically informed masking regimen. We also present two corpora for the Latin language. Acervus contains 24B BERT WordPiece tokens, including over 760K samples from heavily damaged OCR transcriptions, repaired by a ByT5 model trained for Latin text repair. The second dataset, Corpus Trinum, contains a mixture of over 100M strong and weak Latin language semantic triplets. Corpus Trinum is the first Latin semantic triplet dataset, opening the door for Latin language models to be adapted to general sentence semantic similarity tasks. We construct an experimental grid using five tasks and seventeen subtasks, applying a variety of finetuning configurations and subword aggregation techniques for 15 pretrained encoders, 6 LAMB training checkpoints, and an ablation model designed to quantify the impact of different masking regimens. Using this benchmark, we rank Latin encoders and find that LAMB outperforms all other evaluated models on 16 of 17 subtasks, coming within 0.8 percentage points of state of the art on the seventeenth. We provide additional analyses relating Latin encoder performance to tokenizer behavior, syntactic structure, finetuning configuration, pretraining state, subword aggregation, and variation in source texts and domains. In these, we identify unique acquisition dynamics for Latin open class words during terminal decay phases, and establish that for some task categories, finetuning configuration selection can have a greater effect on downstream scores than additional pretraining tokens.

0.27

DraMaze: A Multi-Agent Pipeline with Future-Aware Memory for Diverse Microdrama Script Generation

cs.CL ·2026-10-22 ·Siyang He, Qiqi Wang, Fan Liu, Yuncheng Huang, Pengfei Xu +3 more
benchmark
Abstract

The rapid growth of microdramas has increased demand for high-quality script production. Long-form microdrama script generation with LLMs requires both diverse plot progression and long-range narrative coherence. Our observations reveal that candidate continuations increasingly concentrate on fewer narrative mechanisms as generation progresses, while conventional retrieval may reinforce trope reuse and overlook historical information essential to future plots. To address these challenges, we propose DraMaze, a multi-agent pipeline that explores diverse causal paths through trope memory-guided evolutionary tree search and mutation. It further constructs a Backward Narrative Dependency Tree to organize, retain, and retrieve historical memories based on future narrative goals. We also introduce MiDramaBench, a bilingual benchmark for evaluating script logic and creative quality across different script lengths, together with causal-pattern metrics for narrative repetition and diversity. Experiments show consistent improvements across three backbone LLMs. For 60-episode scripts of up to 60k words, DraMaze reduces causal-pattern repetition by 32.1% and achieves a Logic score of 82.5, outperforming the best-performing baseline by 8.5 points.

0.27

How Ground Truth Shapes Retrieval Scores: Evidence-Aware Evaluation of a Retrieval-Augmented Generation System for Breast Lesions

cs.CL cs.AI ·2026-10-22 ·Miguel Mochizuki Silva, Samuel Ismael, Mateus C. Dias, Thaís Gaudêncio, Yuri Malheiros +1 more
benchmark
Abstract · 18 pages, 3 figures, 6 tables

Purpose: To describe an evidence-aware retrieval evaluation protocol for a clinical retrieval-augmented generation (RAG) system over breast lesion management literature, addressing the dependence of retrieval scores on annotation coverage. Methods: An 18-document corpus and an 18-query development benchmark, with one Yes/No or WH-question per document, were used. Retrieval combined dense embeddings, BM25, and cross-encoder reranking. Chunk sizes of 128 to 1024 tokens, overlaps of 0 to 20%, and four retriever configurations were evaluated. Relevance was defined by evidence spans mapped to all containing chunks, treating overlapping chunks with equivalent evidence as relevant. Retrieval was evaluated using R-Precision, Recall@K, MRR, and nDCG@K. Separately, generation was assessed on a 512-token index at K=1 to 5 using DeepEval claim-level metrics, with deepseek-v4-pro as the LLM judge, and BLEU/ROUGE metrics. Results: The selected hybrid retriever, using 384-token chunks and 20% overlap, retrieved accepted evidence in the top three results for 17 of 18 queries. Recall@3, R-Precision, MRR@3, and nDCG@3 each reached 0.94. Ground-truth curation increased evidence spans from 25 to 37 and accepted chunk targets from 36 to 51 without changing retrieval rankings. Conclusion: Cross-encoder reranking achieved the highest retrieval scores. Expanded relevance annotations increased recognized evidence without changing rankings, highlighting annotation sensitivity. The small development set and retrieval-informed curation require independent validation, including radiologist assessment of evidence sufficiency.

0.27

GateFormer: Forecast-Space Decomposition for Intra-Hour Solar Irradiance Prediction from Sky Images

eess.IV cs.AIcs.CVcs.LG ·2026-10-22 ·David Hamlyn, Sunny Chaudhary, Jonathon Hare, Tasmiat Rahman
publishedbenchmark
Abstract

Sub-hourly dispatch, balancing, and storage control increasingly depend on forecasts that anticipate rapid variability in solar generation. Accurate sky-image-based irradiance forecasting requires a combination of visual cloud dynamics with recent irradiance state while adapting to different short-term irradiance-change regimes. To address this, we introduce GateFormer, a GHI-conditioned dual-decoder ViT for intra-hour global horizontal irradiance (GHI) forecasting through forecast-space decomposition. GateFormer embeds observed GHI into the visual token stream, allowing interaction with sky-image patch features through self-attention. From this shared representation, a Trajectory Head predicts future GHI sequence, while a Delta Head predicts final-horizon change from the latest observed GHI. The two final-horizon forecasts are adaptively weighted in prediction space, with magnitude- and direction-supervision targeting the remaining error. On the NREL-TSI benchmark, GateFormer attains 125.01 W/m^2 RMSE at +30 min, improving over a state-of-the-art transformer baseline by 9.6% under paired-timestamp evaluation. Across NREL-TSI, Chilbolton, and SIRTA, GateFormer achieves the lowest final-horizon RMSE among the evaluated baselines, with forecast skill of 22.97-28.86% relative to smart persistence. Component ablations show adaptive weighting outperforms both individual decoders and fixed averaging, distinguishing GateFormer from conventional forecast combination, while the supervised correction further reduces RMSE by 1.68 W/m^2 and MAE by 4.86 W/m^2. Ramp-regime diagnostics show larger gains during operationally important positive ramps, reducing RMSE by 26.27 W/m^2. These results identify forecast-space decomposition as an alternative forecasting formulation, with potential relevance to other multimodal short-horizon forecasting tasks and deployment using only sky images and measured GHI.

0.26

A General-Purpose Dual-Path Interaction Framework for Multi-View Medical Image Segmentation: Alleviating Data Volume Dependency in Small-Sample Regimes

cs.CV ·2026-10-22 ·Shuhuan Wang, Yuanyuan Cui, Pengju Wang, Huimin Guo, Ling Yue +8 more
Abstract · 20 pages, 5 figures

The performance of medical image segmentation depends on a large amount of annotated data, and most segmentation models experience a significant drop in performance in low-sample-size scenarios. Existing semi-supervised and weakly supervised methods rely on specific network architectures, while traditional multi-view-assisted segmentation methods require strictly paired samples, which limits their clinical applications. To address these issues, we propose DPIF, a general-purpose plug-and-play dual-path interaction framework compatible with arbitrary encoder-decoder segmentation networks, which leverages cross-view anatomical complementarity to reduce annotation reliance. DPIF first performs Sinkhorn optimal transport-based cross-view distribution alignment to establish global semantic correspondence without sample-level pairing. It then adopts conditional mutual information-guided feature decoupling to enable bidirectional knowledge transfer only within the shared semantic space, suppressing view-specific noise. Finally, persistent homology-based topological regularization encodes anatomical invariants as unsupervised priors to mitigate overfitting. Experiments were conducted using 24 baseline models on a private multicenter echocardiography dataset and two public datasets. The results show that DPIF-enhanced models outperform single-view baseline models when trained on comparable amounts of data. Notably, most dual-path variants trained on 40\% - 60\% of the data achieve performance comparable to that of single-view models trained on the full dataset. This demonstrates that DPIF provides a general-purpose and scalable solution for data-scarce multi-view medical image segmentation problems.

0.26

Prompt Sensitivity in Tool-Augmented Model Comparisons

cs.CL cs.AIcs.LG ·2026-10-22 ·Mateusz Modrzejewski, Tomasz Radzikowski
Abstract · 9 pages plus 8 pages of supplementary material; 2 figures and 4 tables in the main text, 2 figures and 5 tables in the supplement

Language models are compared on question-answering benchmarks that pose each question in one wording, and a reworded question can change which of two models scores higher. We ask whether giving the models tools makes such comparisons more stable. An accuracy gain cannot answer this, because tools change several steps between the question and the scored answer at once. We propose a matched protocol that separates those steps: the answer readout, reasoning, the decision to call, the tool's answer, and the report supplied without a call. We apply it to six open models in five domains under five wordings. We find that tools raise accuracy where a program answers the question and widen the wording sensitivity of many pairwise comparisons. The widening follows how often a model calls, and call frequency follows the wording for some models and not others. It also depends on the interface and the instructions: a native tool-calling interface and a second tool description leave no detected widening in molecules, and sensitivity persists where tools return intermediate results. These findings show that deterministic reports do not guarantee stable comparisons. We recommend reporting call frequency and comparison stability under several wordings beside accuracy.

0.26

FilmFlow: Agentic System for 3D Virtual Film Cinematography

cs.CV ·2026-10-22 ·Yiming Zhang, Jiarui Guo, Haojia Wei, Yifei Liu, Yifan Yang +3 more
benchmarkcommunity pick
Abstract

Complex camera motion is central to visual storytelling, yet translating cinematographic intent into controllable, executable camera trajectories remains challenging. We introduce FilmFlow, an agentic framework that converts natural-language instructions or reference videos into explicit camera trajectories for dynamic 3D environments. To generate these trajectories, a Director agent constructs executable camera plans by composing high-level motions from concurrent atomic operations that jointly control translation, rotation, and framing. The Director is supported by three specialized agents: an Interpreter that extracts scene-grounded cinematographic intent, a Scout that selects initial viewpoints, and a Critic that evaluates rendered results and provides visual feedback for iterative plan refinement. We also introduce CinemaBench, comprising 150 Blender scenes from self-created and licensed sources, with five camera instructions per scene for 750 text-directed tasks and 60 reference videos associated with a subset of scenes. CinemaBench supports evaluation of both text-directed camera generation and video reference-directed camera generation. Across a diverse set of evaluated baselines, FilmFlow achieves the strongest benchmark performance and the highest user preference, demonstrating effective spatial reasoning, temporal coordination, and visual composition for executable 3D cinematography.

0.26

CODI: Improving Self-Supervised Vision Transformers with Cross Distillation

cs.CV ·2026-10-22 ·Siran Dai, Qianqian Xu, Peisong Wen, Yang Liu, Xiaochun Cao +1 more
code/data
Abstract · 35 pages, 11 figures, 14 tables

Self-supervised Vision Transformers learn strong image-level representations, but their patch-level representations can be unreliable for dense prediction. We study this mismatch and find that the class token concentrates global semantics on a small number of representative patches, while dense prediction errors often occur on low-attention regions. We provide a local gradient analysis showing that, along the [CLS] attention readout, image-level self-distillation gradients are weighted by [CLS]-to-patch attention mass. To provide masked patches with global semantics without changing the architecture, we propose Cross Distillation (CODI), which mixes a small amount of the teacher class-token distribution into each teacher patch-token target. CODI gives every masked patch an attention-independent semantic correction while preserving the standard self-distillation architecture. Experiments on dense and image-level benchmarks show that CODI improves both dense and image-level performance under aligned pretraining settings. Code is available at https://github.com/EldercatSAM/CODI.

0.26

TutorCaseBench: A Case-Grounded Benchmark for Evaluating LLM Tutors in Supporting Students with Challenging Behaviors

cs.CL cs.HC ·2026-10-22 ·Yeon Su Park, Chanyoung Park, Tae Soo Kim, Seungju Kim, Hyunwoo Kim +3 more
benchmarkcode/data
Abstract · Under Review

LLMs are increasingly used as AI tutors, yet existing benchmarks overlook the challenging behaviors common among students with learning and behavioral difficulties. We introduce TutorCaseBench, a benchmark that grounds AI-tutor evaluation in qualitative case studies from pedagogy, educational psychology, and special education. Our pipeline converts documented student categories into case-specific scenarios, tutor-student dialogues, and rubrics grounded in source evidence. The benchmark contains 923 dialogues across five challenge types, evaluated through open-ended response generation (1,560 items) and multiple-choice selection (2,537 items). Across 11 state-of-the-art LLMs, even the strongest model satisfies a rubric criterion in only 55.8% of open-ended cases, while models are better at selecting appropriate strategies than generating situated responses. These results reveal a gap between intervention recognition and context-specific adaptation, especially for open-weight models. All data are publicly available at https://github.com/kixlab/TutorCaseBench.

0.26

UniVOE: A Removal-Driven Framework for Interaction-Aware Universal Video Object Editing

cs.CV ·2026-10-22 ·Shuzhou Yang, Lanqing Guo, Lingen Li, Weiqi Li, Zhao Zhong +2 more
code/data
Abstract

Video object removal and injection are reciprocal editing operations between object-present and object-absent video states, yet they differ substantially in training difficulty. Removal can be effectively bootstrapped from video inpainting and further adapted to erase object-related interaction effects. Injection, however, requires large-scale paired supervision to model object appearance, dynamics, and scene-dependent effects, while perfectly aligned real-world pairs are difficult to collect and synthetic data lacks sufficient realism and diversity. To address this challenge, we present UniVOE, a removal-driven framework for training both video object removal and injection models. We first train an effect-aware remover by transferring video inpainting priors and finetuning on object-present/object-absent pairs. The trained remover converts real object-present videos into object-absent counterparts, creating large-scale paired supervision for injection. An injection model is trained on these pairs in the reverse direction to recover objects together with their dynamics and interaction effects. Moreover, since pixel-level reconstruction alone cannot adequately capture perceptual realism, we introduce a reward-based post-training method that aligns injection results with human preferences using a proprietary reward model. Experiments against strong open-source and commercial systems demonstrate that UniVOE achieves state-of-the-art performance on both object removal and injection, particularly in challenging interaction-effect cases.

0.26

An Empirical Study of Recent Large Language Models for Knowledge Graph to Text Generation on Standard Benchmarks

cs.CL ·2026-10-22 ·Priyanka, Renaud Delbru, Souhail Bakkali, Mickaël Coustaty, Antoine Doucet
benchmark
Abstract · 18 pages, 6 figures, 17 tables, Preprint, Under review

Knowledge graph-to-text (KG-to-text) generation verbalizes a set of knowledge graph triples into fluent natural language. While fine-tuned encoder-decoder models have long been the standard approach, recent LLMs are becoming increasingly competitive. We benchmark recent LLMs against fine-tuned encoder-decoder models, using reference-based metrics and graph-grounded reference-free LLM-based evaluation validated against human annotations. We observed that recent LLMs substantially narrow, and in several settings close, the gap with fine-tuned models on reference-based metrics, while performing better on graph-grounded metrics. LLM-based evaluation scores show limited correlation with reference-based metrics, suggesting that the two approaches capture different aspects of generation quality. Graph-grounded evaluation reveals that prompted models produce fewer contradictions and triple omissions than fine-tuned models, and these errors increase more rapidly with graph size for fine-tuned models, highlighting differences between the two approaches that reference-based metrics may not fully capture.

0.25

Gate3DGen: A Pre-processing Model that Selectively Purifies Real-World Images for 3D Generation

cs.CV ·2026-10-22 ·Xiaolin Li, Zihao Huang, Fanpeng Meng, Chen Yang, Guanjun Wu +3 more
benchmarkcode/data
Abstract · Project page: https://supra-3d.github.io/gate3dgen

Single-image 3D generation has achieved high fidelity on curated benchmarks, yet degrades sharply on real-world images with uneven illumination, blur, and occlusions. Purifying such inputs requires deciding when to intervene and preserving the visual cues needed for 3D generation, while leaving clean inputs unchanged. We introduce Gate3DGen, a selective pre-processing model that gates image inputs to existing 3D generators without retraining them. A degradation-aware router bypasses inputs classified as clean and extracts features to guide purification of the remaining inputs. Conditioned on these features, a 3D-oriented one-step purifier purifies degraded inputs with spatial losses that encourage preservation of object identity, contours, and textural cues. To provide paired supervision, we construct a synthetic data engine that renders clean views of 3D assets and synthesises composite degradations. On the synthetic benchmark, the purifier improves image fidelity and downstream 3D geometry in only 2.3s on average, reducing TRELLIS.2 mesh distance by 61.5% relative to degraded inputs. VLM evaluation and human preferences on real photographs further support identity preservation and degradation correction. More demos and code are available at https://supra-3d.github.io/gate3dgen.

0.25

Prompting Strategy Effects Depend on the Judge: A Multilingual Study of LLM-as-a-Judge Robustness

cs.CL cs.AIcs.LG ·2026-10-22 ·Asal Mehrabi, Mohammad Aliannejadi, Clemencia Siro
benchmark
Abstract · 23 pages,

LLM-as-a-Judge evaluation is increasingly applied to multilingual tasks, yet prompting strategies such as Chain-of-Thought and Role Conditioning are recommended based solely on English benchmarks. Whether their effectiveness transfers across judge models and languages remains unknown: prompt-sensitivity studies are largely English-centric, while existing multilingual judge studies fix the prompt template. In this work, we address this gap by measuring this interaction through a benchmark of 9 prompting strategies, 4 open-weight judges, 27 languages, and 4 tasks across 13 datasets, yielding over 1M pointwise scores on gold-reference outputs, validated against human judgments. We find that prompting strategies are not transferable: the choice of the judge model dominates the choice of prompting strategy, strategy effects reverse direction across judges, and stability degrades in low-resource languages, where these effects are concealed by aggregate reporting. These findings show that prompting effects in multilingual evaluation are conditioned on the judge, task, and language. Consequently, prompting strategies should be validated for the intended evaluation setting, and cross-study comparisons should account for both judge and prompt configuration.

0.25

How Well Do Decision Models Transfer to Vietnamese? An Empirical Study of Text Classification

cs.CL ·2026-10-22 ·Phat Huu Ha, Nguyen Xuan Phuc, Nong Nhut Quy, Trinh Tran Tran, Nguyen Minh Triet +1 more
Abstract

Recent advances in specialized decision models have expanded the possibilities for zero-shot classification, yet their effectiveness on Vietnamese remains underexplored. This study evaluates four specialized decision models across 13 classification settings spanning six task groups, examining zero-shot performance, sensitivity to label-description language, and robustness to Vietnamese linguistic variations. The evaluation considers English, Vietnamese, and bilingual label descriptions, alongside common linguistic perturbations involving teencode, typographical errors, and diacritic removal. Under English label descriptions, JEV achieves the strongest overall performance, ranking first in 11 classification settings. The effects of label-description language vary across models and datasets, with bilingual descriptions providing no consistent improvement. Robustness experiments reveal that diacritic removal causes the most substantial performance degradation, reducing mean macro-$F_1$ by 9.51--15.47 points across model configurations, considerably more than teencode or typographical errors. These findings highlight the importance of label-description choices and Vietnamese orthographic variation when evaluating specialized decision models for zero-shot text classification.

0.25

Moving Regions, Not Moving Objects: Segmenting What Actually Moves under Camera Motion

cs.CV ·2026-10-22 ·Wenhua Jin, Tiancheng Chen, Sheng Tang, Jiahao Li, Yupeng Liu +3 more
benchmarkcode/data
Abstract · Project page and video demonstrations: https://kosmoresearch.github.io/MoRSe/

Moving object segmentation is conventionally formulated at the sequence and object level, marking an entire instance as moving once it moves anywhere in a video. Recent instantaneous formulations distinguish moving from stationary intervals, but still generate masks at the object level through frozen segmentation foundation models. Because the mask is derived from an instance-level proposal, motion is typically attributed to the entire instance rather than localized to the regions that actually move. Moreover, such instance-level predictions can produce masks even when no part of the instance is actually moving, revealing a more fundamental issue: motion is not an intrinsic property of an instance; it can only be determined by comparing at least two frames. Building on this premise, we introduce MoRSe, which jointly interprets two frames to predict a moving-region mask, distinguishing actual scene motion from apparent motion induced by the camera. A single RGB encoder replaces multi-stage foundation stacks, so no optical flow, point tracks, or separate mask generator is needed at inference. We release MoRSe-5K, a benchmark of 42 sequences with Static, Dynamic-Static, and Dynamic settings designed to evaluate the model's ability to distinguish and localize moving and stationary regions under different camera and scene motion conditions. MoRSe outperforms both specialized moving object segmentation methods and dense 4D reconstruction methods with motion-thresholded predictions, while producing the fewest false positives on sequences where nothing moves.

0.25

LOCI: Localized Continuous Control for Multi-Subject Image Editing

cs.CV ·2026-10-22 ·Sai Vidyaranya Nuthalapati, Shlok Kumar Mishra
Abstract

Recent instruction-following image editors such as Flux.1-Kontext enable text-driven editing but lack fine-grained control over where and how much an edit is applied. Existing approaches provide either continuous but global control or spatially localized editing at a fixed strength. To our knowledge, none jointly supports explicit spatial localization, continuous per-region strength, and independent simultaneous control of multiple subjects. We present LOCI (LOcalized Continuous control for multi- subject Image editing). Given a multi-subject image with per-subject regions and instructions, LOCI edits each subject at an independently chosen, continuously adjustable strength, for instance gradually aging one person while adding a beard to another with minimal cross-region leakage. LOCI pairs subject-level attention binding for spatial isolation with a Regional LoRA layer that replaces the LoRA's scalar coefficient with a per-token scaling vector aligned to the spatial layout. A single adapter is trained at one suppression point, and continuous edit strengths follow by rescaling its regional residual at inference. Experiments on multi-subject scenes demonstrate that LOCI achieves strong edit fidelity with significantly reduced cross-region interference compared to baselines, while maintaining continuous control over edit strength

0.25

AGF-Net: A Robust Multimodal Emotion Recognition Model for Asynchronous Audio-Visual Signals Based on Predictive Dynamic Fusion with Negative Learning

cs.MM cs.SD ·2026-10-22 ·Chuan Han Li, Yun Chien Cheng
hot direction
Abstract

This study develops an audio-visual emotion recognition system that integrates automatic utterance segmentation with multimodal fusion. Unlike conventional methods that rely on manually annotated utterance boundaries, the proposed system automatically generates candidate segments from continuous dialogue and extracts the corresponding audio and visual features for emotion classification. To address the challenges introduced by automatic segmentation, the proposed AGF-Net integrates Audio-Guided Fusion (AGF), Multimodal Negative Learning (MNL), and Predictive Dynamic Fusion (PDF) to improve cross-modal fusion, modality balance, and robustness to varying modality quality. Experimental results show that AGF-Net achieves 64.92% Accuracy and 65.47% F1-score under clean conditions, 62.13% Accuracy and 63.08% F1-score under moderate noise, and 58.19% Accuracy and 59.41% F1-score under higher noise. Overall, the proposed framework reduces reliance on manually annotated utterance boundaries and maintains competitive performance under automatically segmented and noisy audio-visual inputs.

0.25

PixGR: Pixel-Space Generative Decoding for Fast Feed-Forward 3D Reconstruction

cs.CV ·2026-10-22 ·Ranran Huang, Wenjing Bian, Weixun Luo, Ye Mao, Krystian Mikolajczyk
code/data
Abstract · Project page: https://ranrhuang.github.io/pixgr/

We present PixGR, an efficient framework that integrates lightweight pixel-space generative modeling into feed-forward 3D Gaussian reconstruction to model scene uncertainty. Context observations are encoded once into a scene latent representation, which conditions generative decoding from noisy target views to produce a shared 3D Gaussian representation. We formulate this process using pixel-space flow matching, enabling end-to-end optimization of 3D reconstruction with direct pixel-space generative supervision through differentiable rendering. This avoids VAE-based latent compression and error propagation inherent in previous two-stage pipelines. To preserve feed-forward efficiency, PixGR caches the scene representation and performs iterative inference using a lightweight DiT decoder. Experiments demonstrate that PixGR outperforms regression-based approaches in challenging sparse-view settings, particularly with widely spaced observations and view extrapolation. Meanwhile, it maintains sub-second inference and remains competitive with substantially heavier pretrained generative models. Our project page is available at https://ranrhuang.github.io/pixgr/.

0.24

Where Did the Light Come From? Diagnosing Cast-Shadow Reasoning in VLMs and Borrowing the Missing Shadow Prior from Image Generation Models

cs.CV ·2026-10-22 ·Ryugo Morita, Shunsuke Kitada, Stanislav Frolov, Brian Bernhard Moser, Ko Watanabe +3 more
published
Abstract · Accepted to WACV2027

Shadows are observable consequences of latent physical causes, arising from the interaction between illumination, object geometry, and scene structure. A model that understands shadows should be able not only to read a shadow's direction, but also to infer where a shadow should fall when it is absent, and to judge whether a shadow is physically consistent with the scene. We introduce ShadowBench, built from real photographs, which poses these three questions to vision-language models (VLMs). From each scene, we build a correct-shadow, a shadow-removed, and a human-annotated wrong-direction state. Evaluating 19 open VLMs (2B to 78B) and frontier proprietary models, we find that VLMs read a visible shadow's direction well above chance and improve with scale, yet infer the absent direction only weakly and fall to chance at judging physical consistency even under a bias-free two-alternative test that removes the yes-bias of single-image naturalness questions. Only Gemini-3.1-pro exceeds chance on validity, and even then only modestly and far below humans. Chain-of-Thought (CoT) does not help. The missing ingredient is not reasoning but knowledge that image generation models (IGMs) already hold. Routing a VLM through an IGM lifts its inference well above its direct attempt and helps it flag the wrong-direction shadows it otherwise accepts. Cast-shadow direction is a single facet of illumination understanding, but the sharp gap it reveals between VLMs and IGMs suggests image-generation or reconstruction objectives as a route toward illumination-aware VLMs.

0.24

Automated Nucleus Segmentation in Histology Images: An Attention U-Net Approach with Comprehensive Preprocessing and Inference Pipelines

cs.CV ·2026-10-22 ·Akshat G, Divyansh Gupta, Tusar Kanti Mishra
Abstract · 8 pages, 8 figures

Objectives: In the area of digital pathology, nucleus segmentation is an essential step for quantitative analysis; however, high nuclear density and staining heterogeneity in histology images pose a great challenge. Methods: This study implements a lightweight Attention U-Net model on the MoNuSeg 2018 dataset to overcome these challenges. The proposed pipeline integrates comprehensive image and mask preprocessing, including stain deconvolution and multi-pass mask smoothing, with a high-fidelity overlapping patch-stitching strategy during inference on high-resolution images. Results: The model achieved an Intersection over Union (IoU) score of 0.8745 on full-resolution, unseen test images. This result demonstrates the effectiveness of attention mechanisms in enhancing feature localization and validates the stitching methodology for generating seamless and accurate segmentation maps. Conclusion: Attention U-Net is an effective and competitive method for automated, scalable nucleus delineation in challenging clinical contexts.

0.24

HimmBO: High-Dimensional Interactive Model-Merging Bayesian Optimization for Image Generation

cs.CV cs.GR ·2026-10-22 ·Jinyuan Mao, Zhihao Shen, Chenxi Liu
Abstract

Fine-tuning has become a common way to customize text-to-image models, leading to community collections of over 100K lightweight adapters. Adapters sharing a base model can be merged with continuous coefficients. This turns a large collection into a high-dimensional continuous design space, which recent work explores interactively using preferential Bayesian optimization (PBO). However, existing approaches are limited to a few dozen preselected adapters and require expensive image generation in every interaction round. We introduce HimmBO, a two-stage method that scales preference-driven adapter merging to much larger collections while reducing image-generation cost. Its screening stage recovers the active adapter set through level-set estimation over pre-generated images, requiring no image generation. The refinement stage then optimizes merge coefficients in the recovered low-dimensional subspace. Both stages use a style-aware kernel, allowing preference information to transfer between visually similar adapters. In simulated-user image-matching experiments, HimmBO achieves higher final similarity than all comparisons while reaching a given similarity with under one tenth of the image generations required by prior work. Its performance remains stable as the collection grows to 1080 adapters, while competing methods degrade substantially.

0.24

Self-Supervised Multibeam Sonar Restoration via Coherent Forward Modeling

eess.IV cs.CVeess.SP ·2026-10-22 ·Zixin Wang, James Hopgood, Mike Davies
Abstract · 16 pages, 8 figures, includes supplementary material

Multibeam imaging sonar produces intensity images degraded by coherent speckle and system-induced blur, while clean reflectivity targets are rarely available. Existing despeckling methods often rely on simplified intensity-domain noise models, synthetic clean targets, or blind-spot self-supervision, and typically do not jointly model deblurring and speckle suppression. We propose a physics-consistent self-supervised framework for joint despeckling and deblurring of multibeam imaging sonar. A stochastic coherent image-formation model yields a conditional-mean forward operator that relates latent scene reflectivity to expected intensity through the squared magnitude of the system point spread function. We then use registered multi-frame sonar sequences to train without clean ground truth images, showing that, under an idealized inter-frame conditional-mean condition, the expected paired loss equals the corresponding forward-prediction loss up to a network-independent constant. To stabilize the ill-conditioned inverse problem, we further adapt robust equivariant imaging using stochastic coherent simulation and introduce a paired forward prediction error metric for ground-truth-free evaluation. Experiments on two real multibeam sonar datasets demonstrate a favourable balance between forward-prediction agreement, homogeneous-region speckle suppression, and preservation of visible structural boundaries compared with representative baselines.

0.24

PiHand: Latent Hand-Object Interaction Priors for Physics-Based Policy Learning

cs.CV ·2026-10-22 ·Niklas Berndt, Xianghui Xie, Enes Duran, Yixiao Wang, Gerard Pons-Moll
code/data
Abstract · Project page: https://virtualhumans.mpi-inf.mpg.de/PiHand/

Synthesizing physically grounded dexterous hand-object interactions is important for VR/AR, gaming, and robotics, yet remains challenging because subtle contact dynamics, diverse object geometries, and target trajectories must be jointly respected while maintaining natural grasps. Existing methods either track mocap trajectories with limited transfer to novel objects, or rely on hand-only generative priors that fail to capture coupled hand-object motion, sacrificing contact fidelity and sample efficiency. To tackle these challenges, we introduce PiHand, a Physically grounded Interaction-aware Hand motion synthesis framework. PiHand first trains expert policies that track mocap demonstrations in simulation, distills them into a unified hand-object interaction prior using a goal-conditioned VAE, and finally solve diverse dexterous manipulation tasks through bounded residual exploration in its latent space. With only a single reward and no complex curriculum training, this yields general tabletop manipulation policies that generalize to unseen objects and trajectories on GRAB and DexYCB. The same prior transfers to object catching and bimanual handover, two tasks that are rarely seen in its training data. Across all tasks, PiHand consistently outperforms baselines in all metrics. We will release the processed data and model checkpoints.

0.24

Verbalizing Activation Differences with Natural Language Autoencoders

cs.CL ·2026-10-22 ·Andrea Blasi Núñez, Gianluca Barmina, Federico Torrielli, Thiusius Rajeeth Savarimuthu, Peter Schneider-Kamp +1 more
Abstract · Under Review

Activation verbalizers translate the internal representations of large language models into human-readable text. The recently proposed natural language autoencoders (NLAs) learn this by reconstructing activations through a natural language bottleneck. However, it remains unclear how well these methods generalize beyond the setting they were trained for. Here, we study this question for NLAs along four closely related questions: (i) Do NLA verbalizations exhibit compositional structure with respect to their input? (ii) Do NLAs generalize to finetuned versions of the model they were trained for? (iii) Can NLAs recover finetuning objectives in model diffing? (iv) Can NLAs be specialized for model diffing? Across four models from three families, our experiments provide evidence supporting a largely positive answer to all four questions. Specifically, verbalizations of composed analogy activations mostly name the target while rarely mentioning the subtracted word, and an NLA can be reused on lightweight finetunes of its original model without retraining, with reconstruction quality degrading noticeably only for the largest model. Verbalizing activation differences between Taboo model organisms and their original models recovers the concealed secret word for most organisms, and reinforcement learning on activation differences further improves this recovery on held-out organisms. We also find that NLAs can recover the finetuning objective without the finetuned model generating any token, since the most informative token positions precede the user prompt. NLAs do not yet match state-of-the-art model diffing methods on the emergent misalignment detection task, but outperform a blackbox baseline. Our findings suggest that NLAs can meaningfully verbalize differences in activation space, rendering NLAs a promising direction for future work.

0.24

Evaluation of Front-End Feature Matching for Lunar Visual SLAM

cs.CV ·2026-10-22 ·Guillem Casadesus Vila, Grace Gao
benchmark
Abstract · 16 pages, 8 figures, 5 tables. Submitted to the 2027 IEEE Aerospace Conference

Lunar rovers will primarily rely on cameras for localization until navigation satellite services are deployed, and will continue to need them afterwards for relative navigation, hazard detection, and surface operations. Visual odometry (VO) and Simultaneous Localization and Mapping (SLAM) mostly estimate motion by matching features between frames. However, feature matchers are developed on urban, indoor and driving scenes, planetary studies compare them on orbital registration or within complete SLAM systems, and real lunar imagery with reference poses is scarce, so matching between rover frames on low-texture, harshly lit terrain, and its effect on VO, has not been compared. We benchmark eleven classical and learned detector-matcher combinations on nine synthetic, lunar-testbed, and terrestrial-analogue datasets, and a feature tracker, scoring pairwise pose accuracy against baseline and the trajectory error of the resulting VO. Under a common rotation-only metric, the SuperPoint pairings outperform SIFT with ratio-tested FLANN matching on both lunar testbeds and one analogue site, with smaller or unresolved differences on synthetic data. When rotation and translation direction are scored jointly, this SIFT pairing exceeds both SuperPoint pairings on four synthetic datasets, so synthetic results alone do not establish relative performance on real rover imagery. In stereo VO, the SuperPoint pairings are not consistently more accurate than SIFT, but only they keep tracking on all eight sequential datasets. Pairings with classical detectors depend on lighting while SuperPoint changes little: a low spotlight makes them more accurate than diffuse light on the same testbed views, consistent with more contrast and detected keypoints. SuperPoint with LightGlue is the fastest learned combination (16 ms per frame), and DINOv2 retrieval proposals still need geometric verification.

0.24

Cross-Spectral Feature Fusion for Multispectral Document Binarization

cs.CV eess.SP ·2026-10-22 ·Yasas Savinda, Maheshi Buddhinee Dissanayake
unusual
Abstract

Historical document binarization remains challenging when faded ink, bleed-through, and background degradation exhibit wavelength-dependent behavior across visible and near-infrared observations. This study investigates whether adaptive cross-spectral feature fusion and targeted structural guidance can improve multispectral document binarization beyond fixed fusion and generic attention mechanisms. Four registered SMADI bands at 500, 600, 700, and 1000 nm are processed using a multi-branch conditional generative adversarial network (cGAN). Visible, NIR, and structural feature streams are fused at encoder depth L = 3 using spatial- and channel-wise softmax gating, while auxiliary edge maps derived from selected VIS-NIR band pairs provide targeted structural guidance. Nine controlled fusion and structural configurations are evaluated using five-fold document-level cross-validation, DIBCO-standard metrics, and Holm-corrected statistical testing. The F3 + NIR edge-guided configuration provides the strongest overall performance among the evaluated models, achieving mean F1 = 0.7916, p-FM = 0.8044, and IoU = 0.6654. Relative to the fixed Mid AVG baseline, it improves F1 by 0.0269, p-FM by 0.0360, and IoU by 0.0321, with all three gains remaining statistically significant after Holm correction. Mean NRM and DRD are also reduced, although these reductions are not statistically significant after correction. Targeted cross-spectral structural guidance provides an effective mechanism for regulating intermediate multispectral fusion, improving primary foreground-overlap performance while retaining the complete four-band observation.

0.24

TaxoSAE: Rethinking Sparse Autoencoders through Taxonomic Logic

cs.CL ·2026-10-22 ·Xinyu Guo, Minglai Yang, Clayton T. Morrison, Eduardo Blanco, Mihai Surdeanu
Abstract

Recent Sparse Autoencoders (SAEs) decompose the representations of large language models (LLMs) into interpretable features, enabling precise readout and control of encoded knowledge. However, existing SAEs treat each feature as a standalone unit, whereas real-world knowledge is often taxonomically organized. As a result, readout carries no guarantee of taxonomic consistency, and control lacks the granularity to specify the scope of an edit. We propose TaxoSAE, which explicitly encodes taxonomical information, built on is-a (hypernym-hyponym) relations, into the SAE feature space. TaxoSAE mirrors a taxonomy with a stack of SAEs, one per taxonomy level, and introduces a taxonomic alignment loss that trains each concept to be active whenever one of its hyponyms is. On two factual QA taxonomies, one over biographical relations and one over animals, TaxoSAE enables (1) more accurate readout through the taxonomy with concentrated activations, e.g., +31.3 points in reading out the hypernyms of abstract concepts never queried in training; (2) more precise control, with +25.5 points in swap success, while suppressing a concept removes only its own answers (0-2% accuracy vs. 91-100% on other concepts); and, more importantly, (3) readout of unseen concepts through their hypernyms, with up to +37.9 points in hypernym readout accuracy.

0.23

DRAFT: Decoding and Reconstructing 3D Room Layouts from fMRI and Camera Trajectories

cs.CV cs.LG ·2026-10-22 ·Chi-Wei Lee, An-Kai Chen, Cheng-Han Chiang, Yi-Chun Chang, Po-Chih Kuo
Abstract · 42 pages, 7 figures, 18 tables

While functional magnetic resonance imaging (fMRI) decoders reconstruct images and 3D objects, recovering the metric geometry of rooms from a moving viewpoint remains an open challenge. We introduce, to our knowledge for the first time, the task of reconstructing 3D layouts of held-out scenes using only the known camera trajectories and the fMRI signals of participants watching indoor walkthroughs, without any visual frames. The task is hard because of the temporal mismatch between rapid visual changes and blurred fMRI signals. Our method, DRAFT (Decoding and Reconstructing 3D room lAyouts from fMRI and Trajectories), extracts a compact layout representation from the early visual cortex and fuses this neural code with the geometric free space derived from camera motion. To evaluate it, we propose layout r, the Pearson correlation between predicted and ray-cast log depths. On ten held-out scenes (20 participants), DRAFT reaches a layout r of 0.480, outperforming the camera path alone (0.446) and seven fMRI decoders (at most 0.429). DRAFT also works as a plug-in that improves all seven decoders without retraining, raising their layout r from 0.171-0.429 to 0.487-0.493. Brain activity and camera motion can thus recover the metric layout of unseen rooms, a first step toward reconstructing the 3D spaces people see.

0.23

WildFakeToolBench: Benchmarking AI-Generated Image Detectors with Web-Sourced Image Processing

cs.CV cs.CR ·2026-10-22 ·Mingjie Li, Yukun Jiang, Yicong Tan, Yang Zhang
benchmark
Abstract

As AI generated images are getting popular these days, discussions on such generations are widely existed in current public forums, software repositories, and photography tutorials. Moreover, with the help of LLM-assisted code translation, malicious actors can adopt such information to evade detectors without specialized knowledge or local GPU execution of the resulting tools. To evaluate such risks, we introduce \textsc{WildFakeToolBench}, a benchmark suite of $156$ executable image-processing tools derived from $220$ public sources across $142$ domains. Each tool retains its source evidence and passes syntax, execution, and output-distortion checks without using feedback from the evaluated detectors. On $105$ generated images from five generators, we evaluate six state-of-the-art detectors with images processed by the tools in our collected bench to assess their effectiveness against such in-the-wild risks. The strongest individual tools achieve attack success rates of $25.25\%$--$100\%$ across detectors, exceeding the strongest of six selected image-processing baselines for each detector. Evaluating subsets of an eight-tool pool yields worst-case coverage above $80\%$ for every detector. Effectiveness is sparse and detector-specific: high coverage across candidates does not imply reliable one-shot evasion. These findings not only highlight the potential of such web-sourced risks, but also show the potential on establishing AI-image detector robustness evaluation framework based on those in-the-wild information.

0.23

CrisiLangs: A Multilingual Resource for Crisis Warning Generation

cs.CL ·2026-10-22 ·Giacomo Gonella, Stefano Menini, Marco Guerini
Abstract

Communicating with the public during an emergency entails issuing warning messages promptly. In Europe, it often requires producing them in more than one language. Recent work has shown that language models can be trained to generate warning messages as a crisis unfolds, but the available data covers only English. To fill this gap, we present CrisiLangs, a multilingual extension of CrisiText (an English dataset of crisis events paired with warning messages written according to expert communication guidelines) to Italian, Bulgarian, German, and Greek. Expert native speakers translated the warning messages of 1,000 events drawn from 411 scenarios, for a total of 5,000 event-message pairs across five languages. We test the resource to fine-tune a multilingual model on all five languages jointly, with and without the emergency guidelines in the prompt, and compare it with a zero-shot baseline. The fine-tuned models outperform the baseline on ROUGE-L, BLEU and BERTScore in every language and configuration. In a human preference study, low decoding temperatures receive the most preferences in all five languages and in both configurations, while the effect of the guidelines varies by language.

0.23

SPECMEM: Accelerating Agentic Tool Calling via Live Memory Management

cs.CL ·2026-10-22 ·Zihe Song, Wenting Zhao, Ye Liu, Yuanjie Zhu, Ke Xu +2 more
Abstract

Tool-calling agents are mostly optimized for accuracy; in deployment, however, their growing cost is latency: every tool call is decoded autoregressively. Speculative decoding accelerates this without changing the output, and retrieved past calls make strong drafts under real use, where each user's behavior tends to repeat. Yet existing systems keep no cross-session memory, or build the draft store once and never update it. We propose SpecMem, a tool-call memory kept alive and managed throughout deployment: verified calls are written back online into bounded per-user stores with LRU eviction, and a side-effect gate defers irreversible calls until verification confirms. On multi-session, multi-user query streams served by three open-weight models, a frozen store steadily loses acceptance while SpecMem holds it: a 27% gap in mean accepted tokens (+11% to +34% across benchmarks, tracking how much novel work arrives after the store is built) and up to 1.74x wall-clock speedup. Ablations attribute the gain almost entirely to keeping the store live; local per-user scoping adds no acceptance but closes a measured cross-user leakage channel, and gating on declared side effects, rather than on retrieval confidence, is what eliminates irreversible mistakes.

0.23

Rethinking Feature Perturbation for Single-Source Domain Generalization: A Representation-Space Analysis and Spectral Case Study

cs.CV ·2026-10-22 ·Badr-Eddine Benelmostafa, Sohaib Baroud
Abstract

Feature perturbation is widely used in single-source domain generalization (SSDG), yet it is typically evaluated by task performance rather than by the representation changes it induces. We investigate whether perturbations can be compared through measurable descriptors of these changes. We introduce a four-dimensional representation-space description comprising feature displacement $D_F$, relational displacement $D_R$, a task-region ratio $Θ_{\mathrm{Task}}$, and subspace deviation $Θ_{\mathrm{Subspace}}$. We use these descriptors as an empirical analysis framework, not as a universal geometry, to characterize existing perturbation mechanisms. Across two object-detection settings involving style and weather shifts, we observe distinct geometric profiles and a concentration of relatively strong out-of-domain (OOD) configurations within a bounded region of the descriptor space. These findings do not establish a universal optimum but motivate investigating whether regions associated with stronger SSDG performance can be identified without target-domain access. As a case study, we introduce Dual-Invariant Spectral Perturbation (DISP), a non-learned Fourier-phase perturbation with explicit controls over phase magnitude $ρ$, channel coupling $κ$, and frequency support $\mathcal{M}$. We empirically examine how these controls affect the proposed descriptors. DISP achieves strong OOD performance in our experiments, illustrating how explicit perturbation controls can be analyzed through a representation-space description. We hope this viewpoint can contribute to more interpretable, controllable, and ultimately testable approaches to feature perturbation in SSDG.

0.23

PhyEvolve: Multi-Agent State-Evolution Planning for Physics-Aware Video Generation

cs.CV ·2026-10-22 ·Yang-Hao Zhou, Lixing Tan, Zhiqi Li, Guang Yang, Xiao Yang
Abstract

Recent advances in text-to-video generation have substantially improved visual fidelity, yet existing models still struggle to understand the physical evolution implicit in user prompts, making it difficult to generate videos with continuous physical state evolution. Here, we propose PhyEvolve, a multi-agent framework that enhances natural-language prompts through structured state-evolution planning. PhyEvolve consists of three agents: a Director, a Physical Evolver, and a Composer. Given a user prompt and an optional first frame, the Director grounds the scene and disentangles physical events from observation changes. The Physical Evolver uses scene evidence to identify physical mechanisms then constructs causally linked state transitions that unfold throughout the video, whether visible or not. The Composer utilizes these transitions to enhance the original prompts. Without retraining or model modification, PhyEvolve works directly with existing generators through prompt replacement. Extensive experiments on four major benchmarks show that PhyEvolve achieves state-of-the-art performance, outperforming the cutting-edge prior agentic baseline by 7.9 percentage points on average. These results demonstrate that explicit causal state-evolution planning improves physical consistency and state continuity in generated videos through natural-language prompt enhancement.

0.23

A Reality Check of Inference-Time Context Augmentation Methods

cs.CL ·2026-10-22 ·Haeun Yu, Geert Heyman, Pepa Atanasova, Isabelle Augenstein
Abstract · 19 pages, 8 figures

Large language models (LLMs) often fail to utilise the context given in the input, resulting in hallucinations or a fallback to parametric knowledge. Context augmentation methods aim to improve model performance by modifying the input, from inserting filler tokens that add computational time to adding relevant content such as elaboration of a context during inference. However, these methods have been evaluated in isolation and only by their performance gains, leaving it unclear how they compare and whether they reliably improve performance. In this paper, we present an evaluation framework spanning five methods that differ in how informative the inserted content is and apply them at the sentence level across seven models spanning multiple sizes on three datasets. We find that accuracy gains are not always genuine improvements but undirected perturbations: the methods correct some predictions while breaking others, particularly when the inserted text allows only for more computational time. Our analysis reveals that methods that bring no additional content primarily change model predictions near the decision boundary. Our findings highlight the need to evaluate the reliability of the accuracy gains before adopting them in practice.

0.23

The Geometry of Metaphor: Probing and Steering Metaphor Representations in Large Language Models

cs.CL cs.AI ·2026-10-22 ·Shihui Li, Mette Andersen, Martha Lewis
Abstract · BlackboxNLP 2026: The 9th Workshop on Analyzing and Interpreting Neural Networks for NLP

Metaphor is a crucial aspect of human cognition. As such, we would like to understand how it is represented in large language models (LLMs) and how it might be controlled. We use mechanistic interpretability techniques to identify causal subspaces in a range of LLMs covering different architectures and sizes. We show that we are able to steer output on classification tasks across different parts of speech. We are also able to steer output on generation tasks, indicating that these techniques can be used to control the metaphoricity of LLM output.

0.23

Query-Time Calibration of CLIP Features for Open-Vocabulary 3D Segmentation

cs.CV cs.AI ·2026-10-22 ·Dominik Engel, Yousef Rajeh, Leon Sick, Ivan Viola
Abstract

Lifting vision-language features such as CLIP into 3D Gaussian reconstructions enables open-vocabulary segmentation through natural-language queries. Although CLIP represents a broad range of semantic concepts, the lifted features of individual scenes can exhibit strong directional concentration, while raw cosine similarity does not account for the scene-specific feature distribution. We show that substantial segmentation gains can be obtained by changing how a fixed lifted feature field is compared with text. Our method partially subtracts a shared scene direction from visual features and text prototypes and whitens text prototypes using a fixed reference vocabulary of generic terms. It then uses soft clustering for multiclass prediction and CSLS with a relevance margin against the reference vocabulary for single-query prediction. Beyond reducing directional concentration, scene centering improves class-relative feature separation in every evaluated scene. Across seven lifting methods and four datasets, the resulting query-time calibration consistently improves mean intersection over union in both multiclass and single-query segmentation. These gains require no retraining, source-image re-encoding, or feature re-lifting, demonstrating the value of scene-conditioned comparison for making better use of existing lifted representations.

0.23

MQAScore: Fine-Grained Text-Music Alignment Evaluation via Large Audio Language Model

cs.SD cs.AIeess.AS ·2026-10-22 ·Wei-Jaw Lee, Chun-Ping Wang, Xuanjun Chen, Yi-Hsuan Yang
Abstract · Submitted to ICASSP 2027

Text-to-music alignment is commonly evaluated by the global similarity between text and audio embeddings, which collapses all described attributes into a single score and obscures which ones are missing. We instead introduce MQAScore, which verifies each atomic musical concept along five clip-level dimensions (genre, instrument, mood/theme, vocal, and usage) by the probability that a large audio language model (LALM) answers affirmatively to a binary question. To assess evaluator reliability, we construct verification labels from presence/absence annotations and LLM-guided concept substitutions, and propose the Sensitivity-Invariance Ratio, which measures how selectively an evaluator responds to meaning-changing substitutions. Experiments show that concept-level scoring improves both question-answering and similarity-based evaluators, and MQAScore can be a more reliable and interpretable fine-grained complement to global similarity for evaluating text-to-music alignment.

0.23

ParaLin++: Accelerating Parallel Diffusion Integrator via Intrinsic Partially Linear Structure

cs.CV ·2026-10-22 ·Jianrong Lu, Zhiyu Zhu, Hui Liu, Junhui Hou
code/data
Abstract

Training-free few-step samplers are often described as nearly quality-lossless under conventional FID/CLIP-style evaluation. It remains unclear whether that conclusion transfers to more realistic and complex generation-quality benchmarks. We first systematize the quality metrics used to evaluate training-free sampling acceleration for diffusion and flow-matching models. We find that representative training-free step-shortening methods often remain nearly lossless under FID/CLIP-class metrics, yet can suffer substantial quality degradation when evaluated with semantic-composition, dense-instruction, and long-text-rendering benchmarks such as GenEval, DPG-Bench, and LongTextBench. Parallel sampling is a notable exception: by retaining the full-step discretization and removing only the solver's temporal dependency, it can preserve near-lossless quality while accelerating sampling. Building on this observation, we propose \textit{ParaLin++}, a step-parity parallel sampler that preserves the full-step discretization while accelerating the solver iterations. Experiments across Stable Diffusion-v2, LSUN Church, FLUX.1-dev, and Ming-Flash-Omni 2.0 show that ParaLin++ preserves the full-step Euler quality profile at $2.6\times$--$4.8\times$, $1.68\times$ on FLUX, and $1.21\times$ on Ming. Project page: https://paralinplus.github.io/.

0.22

Reinforcing intra-operative tumour localisation with weak supervision from pre-operative appearances

cs.CV ·2026-10-22 ·Sidhart Sami, Ali Sharjeel, Sri Atluri, Miao Zhao, Caroline Roney +2 more
Abstract

Limited tumour visibility on intra-operative images complicates surgical navigation, including liver hepatocellular carcinoma (HCC) targeting on computed tomography (CT) imaging. This makes labelled datasets difficult to curate for deep learning, and transferring pre-operative delineations through registration remains challenging due to intra-operative deformation and lack of paired pre- and intra-operative data. We propose to transfer tumour appearances learned from pre-operative magnetic resonance (MR) images to guide localisation on intra-operative CT through weak supervision. A classifier is trained to classify tumour presence on pre-operative images. This is applied to windows on intra-operative images, and despite the domain gap, it assigns higher tumour-presence probabilities on average to tumour-containing windows. These weak cross-domain signals guide a reinforcement learning agent to localise tumours on intra-operative images. We compare direct transfer, few-shot adaptation and synthetic transfer (trained using unpaired MR and CT), which translates CT into synthetic MR before localisation. On an external HCC cohort, synthetic transfer outperformed other tested methods and CT-only baselines. Our framework enables cross-modality tumour localisation without paired MR-CT training images.

0.22

Investigating Multi-Cultural Value Selection Biases in LLMs

cs.CL ·2026-10-22 ·Siddhesh Pawar, Arnav Arora, Dustin Wright, Dirk Hovy, Isabelle Augenstein
Abstract · 27 pages, 13 figures, 18 tables

Large language models (LLMs) serve users whose cultural identities are multi-layered, yet cultural evaluations commonly reduce identity to one label. In this work, we study LLMs under simultaneous identity cues, with a country persona in the system prompt and a second identity signaled in the user turn, either by naming a country (explicit cue) or through the query language (implicit cue). We test World Values Survey (WVS) multiple-choice probing and open-ended sociopolitical generation. In multiple-choice probing, language cues shift answers about three times more than explicit cues but away from the associated country's distribution, and explicit cues do not consistently move answers toward the named country. Across 1.6 million open-ended responses, naming a country closes 30-44% of the gap to the model's own responses for that country, and explicit cues produce more differentiated value changes than language cues. Our results suggest that country and language cues should be evaluated as distinct signals rather than interchangeable proxies for culture.

0.22

From Signals to Evidence: Interpretable Underwater Acoustic Target Sensing with Large Language Models - Part II: Passive Detection

cs.SD ·2026-10-22 ·Zhiyuan Li, Yuge Zhang, Jie Chen, Xiruo Su, Zhengqiao Zhao +4 more
Abstract · 15 pages, 5 figures

Underwater acoustic passive detection is essential to maritime surveillance and intelligent ocean sensing. Existing detection systems typically collapse multichannel array observations into a binary decision or a scalar confidence score, offering limited insight into the physical evidence that supports or contradicts a target declaration. This limitation is acute at low signal-to-noise ratio (SNR), where isolated spectral peaks and transient beamforming responses are inherently ambiguous and physically verifiable targets are characterized by an ensemble of coupled reinforcing cues. Building on the unified from-signals-to-evidence framework introduced in the companion Part I, this paper instantiates the framework for passive target detection. A physics-aware multichannel acoustic encoder encodes the array observations into complementary acoustic evidence tokens through a direction-guided architecture that combines minimum variance distortionless response (MVDR) processing with time-frequency and fine-frequency representations. These tokens are combined with environmental priors and fed into a pretrained large language model (LLM) that evaluates target-present and noise-only hypotheses and generates a structured detection report detailing target presence, bearing, spectral-line locations, received SNR, range, and per-field confidence scores. A physics-parameterized data engine extended to realistic oceanic channels provides traceable supervision at both discrete-label and structured-description levels. Experiments under white-noise and mixed ocean-background conditions demonstrate reliable detection, accurate bearing estimation, and physically interpretable structured reports, validating the feasibility of evidence-grounded underwater acoustic detection.

0.22

Chunk Recomputation under Real Document Revisions: Measuring Update Cost and Retrieval Quality of Fixed, Recursive, and Content-Defined Chunking

cs.CL cs.IR ·2026-10-22 ·Yuning Bian, Lele Xu, Yifan Yuan, Xinyu Yin, Chengyan Zhan
published
Abstract · 6 pages, 3 figures, 3 tables. Accepted at AIBDCC 2026

Chunking strategies for retrieval-augmented generation are almost always evaluated on a static snapshot, yet the documents underlying a knowledge base continue to change. Because chunks are embedded independently, a boundary that moves after an edit forces unchanged text to be re-embedded. We measure this update cost on 120 pairs of adjacent Chinese and English Wikipedia revisions for three stateless chunkers (non-overlapping fixed-size, recursive paragraph-first, and content-defined chunking) under a strict byte-identical reuse rule and a frozen dense encoder. On the 116-article common-success pool, with frozen, length-calibrated parameters, fixed-size chunking recomputes 57-68% of its full new-version embedding-input workload, recursive chunking 38-56%, and content-defined chunking 43-54% (it failed on 4 of 62 Chinese articles), while at these parameters mean evidence recall under a fixed 2048-token retrieval budget differs by less than two percentage points within either language. A cost-only parameter grid shows that the ranking of the two content-aware chunkers changes with chunk-length settings, whereas non-overlapping fixed-size chunking has the highest recomputation cost in every length-matched condition. Controlled edits and a verbatim-presence diagnostic support boundary drift as one contributor to recomputation of text already present in the previous version. Update cost is therefore a substantial, parameter-sensitive dimension that static retrieval evaluations do not expose and that should be reported alongside retrieval quality.

0.22

Spectral Smoothing and Spiking Perturbations Reconfigure Capability Profiles and Selective Task Performance in Large Language Models

cs.CL cs.AI ·2026-10-22 ·Qince Luan, Jianhui Xu, Yang Xu
benchmark
Abstract

Singular-value spectra encode important structure in transformer feed-forward projections, yet their causal role in downstream behavior remains underexplored. We study post-hoc spectral smoothing and spiking interventions on the up_proj and down_proj matrices. Positive perturbation stretches deviations from the geometric mean, making large singular values larger and small ones smaller (spiking), and vice versa. The intervention preserves singular vectors and the geometric mean of the singular values. Across five open Llama and Qwen instruction models, we evaluate the perturbed systems on capability-labeled subtasks, standard and chain-of-thought reasoning benchmarks, judge-scored truthfulness, hallucination, and safety tasks, and TRAIT personality probes. To interpret score shifts, we fit responses in a three-dimensional capability basis consisting of factual knowledge, language understanding, and deductive reasoning. The induced changes suggest structured rather than purely random sensitivity: large positive perturbations more reliably suppress language and deductive responses, moderate negative perturbations can sometimes improve factual or task-local performance on the evaluated grids, and sensitivity varies strongly by model family, benchmark module, and underlying layerwise spectral inequality. These results position spectral perturbation as a lightweight intervention for probing how capabilities are organized and traded off in large language models.

0.22

A Task-Centric Survey of Long-Document Visual Question Answering: Taxonomy, Evolution, and Future Directions

cs.CV cs.CLcs.IR ·2026-10-22 ·Yufei Tong, Guanjie Cheng, Tianle Wu, Leilei Zhang, Hang Zheng +3 more
survey
Abstract · 26 pages, 9 figures, 6 tables

Visual document understanding is essential for acquiring knowledge from visually rich documents and is attracting growing research interest. Existing surveys emphasize architectures or technical approaches, with limited attention to how task bottlenecks shape technical choices and research priorities. We present a task-centric survey that decomposes long-document visual question answering (VQA) into four components: document parsing, evidence organization and retrieval, generator organization and learning, and evaluation. For each, we develop a fine-grained taxonomy and trace motivations, technical responses, and cross-component connections. We focus on explicit retrieval while covering direct reading. Our temporal analysis traces expanding priorities from multimodal content understanding to sufficient, high-fidelity evidence acquisition and sustained evidence management. We outline future directions in evidence sufficiency, verifiable and correctable understanding, and budget-aware coordination, with an outlook on continual document understanding.

0.22

SyRA-CS: Synthetic-to-Real Adaptation for Code-Switched ASR with Meta-Learning

cs.CL ·2026-10-22 ·Fadel Thior, Ahmer Tabassum, Haania Siddiqui, Hanan Aldarmaki, Thamar Solorio
Abstract

Code-switched (CS) automatic speech recognition (ASR) remains difficult because real CS speech is scarce and synthetic speech does not fully match it. We investigate whether synthetic CS speech is more useful for learning an adaptation-friendly initialization than for ordinary continual fine-tuning. We propose SyRA-CS, a pair-specific synthetic-to-real adaptation framework based on Reptile meta-learning. For each language pair and pretrained model, we meta-train on the corresponding synthetic CS-FLEURS subset and then adapt using one hour of labeled real target speech. We evaluate four architectural settings: Encoder-CTC, FastConformer-RNN-T, Attention Encoder-Decoder (AED) and Speech-LLM. Experiments cover on Arabic-, French-, German-, and Hindi-English, against synthetic-only transfer, real-only fine-tuning, and continual fine-tuning. SyRA-CS improves over continual fine-tuning on WER in all 16 model-pair comparisons and on CER in 15 of 16, with relative macro WER reductions of 17.7% for XLS-R and 5.3-5.8% for the other three models. The gains hold across the tested models but vary in size across architectures, language pairs and metrics.

0.21

From Perpetrators' Digital Footprints to Behavioral Communities: An Ontology-Driven Framework for Bengali Cyberbullying Analysis--- FULL VERSION

cs.CL ·2026-10-22 ·Md. Nesarul Hoque, Md Hanif Seddiqui, Miskatul Anwar, Rudra Pratap Deb Nath
Abstract · 32 pages, 8 tables, 13 figures

Cyberbullying involves behavioral and contextual patterns beyond isolated abusive messages. However, existing approaches remain fragmented across content detection, semantic representation, and network analysis, with limited interpretable perpetrator representations, multi-scale structural modeling, and explanations of behavior-driven similarity and influence propagation. These limitations hinder the identification of persistent perpetrators, characterization of behavioral communities, and assessment of potentially influential users. This paper proposes Ontology-Driven Cyberbullying Behavior and Network Analysis (ODCBNA), a perpetrator-centric framework integrating requirements-driven ontology design, knowledge graph construction, behavioral modeling, community analysis, and influence modeling. Using the Bengali Contextual and Diverse Cyberbullying (BenCoDiC) corpus as a use case, ODCBNA identifies $1{,}063$ users with at least three bullying comments or replies as perpetrators and represents each through a $64$-dimensional behavioral and contextual feature vector. Cosine similarity and $k$-nearest-neighbor modeling produce a weighted network with $27{,}554$ edges. Louvain analysis identifies six communities with a modularity of approximately $0.495$ and $27$ sub-communities. Feature-contribution analysis identifies the Video, Impact, and User Profile groups as the strongest contributors to behavioral similarity and community separation, while contextual dimensions support interpretation. Community-constrained influence analysis shows that IM Greedy achieves small positive gains in all six communities, although different seed-selection methods produce comparable diffusion spreads despite limited seed overlap. These findings provide an interpretable basis for understanding perpetrator behavior, community organization, and potential influence patterns in Bengali online environments.

0.21

CoWriter: Guiding Methodological and Synergetic Human-LLM Co-Authoring

cs.CL cs.AIcs.HC ·2026-10-22 ·Shmuel Amar, Benjamin Lieber, Ori Shapira, Debanjana Kar, Sebastian Maximilian Dennerlein +2 more
Abstract

While LLMs have transformed text generation, standard chat-and-canvas interfaces often reduce human authoring to quite passive supervision, causing review fatigue, over-reliance on AI, and the sidelining of the human role in long-form, source-grounded writing. Prior systems typically address isolated writing phases, leaving comprehensive end-to-end human-LLM collaboration largely unexplored. In this paper, we introduce CoWriter, a proactive agentic scheme that adapts expert human (co-)authoring methodologies from cognitive science to flexibly steer the entire writing workflow while actively encouraging the user to engage. In an exploratory user study, compared to a standard LLM baseline, authors using CoWriter better expressed their perspectives and maintained a higher sense of document ownership, supported quantitatively by $2.2\times$ more successful agent engagements. Furthermore, external reader judges rated the resulting texts higher in "human touch" qualities like storytelling and integrating unique personal experiences. Overall, our results demonstrate that proactively engaging users through established human practices is beneficial for effective LLM-supported co-authoring, providing useful methodologies, techniques and insights for future research on collaborative agents.

0.21

SAGE: A hybrid vision-language and geometry framework for work-zone pathway prediction under temporary traffic control

cs.CV cs.RO ·2026-10-22 ·Jiaxi Liu, Keshu Wu, Weizhe Tang, Bofeng Cao, Sikai Chen +3 more
code/data
Abstract · 26 pages, 9 figures, 9 tables. Project page: https://hhjj233.github.io/SAGE-page/

Temporary traffic control redirects vehicles through work zones, requiring automated vehicles to connect an interpretation of the visible devices to a continuous passing path. Geometric predictors provide several plausible paths, yet selecting the one consistent with the temporary control remains difficult. We propose SAGE (Semantic-to-Action Grounded Execution), a read-decide-execute-check framework that separates path construction, path preference, and control compliance within a shared candidate space. The central idea is to use a vision-language model's (VLM's) trajectory as a preference over intact geometric candidates. A role-decoupled generator constructs the candidates, and rank fusion combines each candidate's geometric score with its agreement with the VLM trajectory. Supervision structured around the Manual on Uniform Traffic Control Devices (MUTCD) trains SAGE-VLM to state the lane status and hazard side before its passage decision and trajectory. The predicted hazard side enters image-space predicates derived from MUTCD provisions, which support a final intervention bounded in score and displacement. On 2,306 real-world ROADWork validation episodes, the candidates include a path within 10 px of the human demonstration in 87% of episodes, yet selection regret accounts for 53% of the geometric ranker's error. SAGE reduces average and final displacement errors by 8.3% and 12.0% relative to the ranker and hard-constraint violations by 37%. The trajectory opinion improves all four tested generator-selector combinations. SAGE is comparable to state-of-the-art VLM modules in accuracy, with a faster-responding VLM, and outperforms four other adapted driving VLMs in displacement error and safety measures. Project page: https://hhjj233.github.io/SAGE-page/.

0.21

CoRE: Relating Fine-Grained Chest X-ray Findings in Anatomical Context

eess.IV ·2026-10-22 ·Minjun Kim, Jong Hak Moon
Abstract · 18 pages, 3 figures

Chest X-ray interpretation depends on both the anatomical location of findings and their relationships with other abnormalities. However, relations defined over global finding labels do not explicitly distinguish the same finding across anatomical locations. We propose CoRE(Co-occurrence-guided Region Evidence), a framework that incorporates relationships among findings and anatomical context into fine-grained chest X-ray prediction. It groups anatomical region-finding labels by co-occurrence and uses these groups to guide shared representation learning through visual prompt tuning on a frozen vision transformer. The predictions are supplemented by an anatomically supervised region-evidence stream through region-matched logit residuals. Experiments on MIMIC-CXR with Chest ImaGenome annotations show that CoRE improves region-finding classification over standard and relation-aware prompt-based baselines. Quantitative evaluations on MS-CXR and the Chest ImaGenome gold standard show improved finding localization, supported by qualitative comparisons. CoRE also improves finding-level classification on mapped finding labels in external VinDr-CXR and CheXpert datasets without target-domain adaptation. These results demonstrate the value of jointly modeling finding relationships and anatomical context for fine-grained chest X-ray analysis.

0.21

EL-DIP: A Deep Image Prior Framework for Accurate,Efficient, and Noise-Robust Elastography

eess.IV cs.CE ·2026-10-22 ·Hao Liu, Fuyi Li, Dong Liu, Yue Mei, Sheng Mao
Abstract

Elastography is a key technique for non-invasive imaging, yet it remains severely ill-posed and sensitive to measurement noise, causing blurring and artifacts. We propose EL-DIP, a training-data-free framework based on deep image prior (DIP) that reconstructs heterogeneous shear modulus distributions. Inverse distance weighting interpolation enables arbitrary-shaped domains, and adjoint sensitivity analysis provides efficient gradient computation. Numerical and experimental studies show that EL-DIP accurately recovers inclusion geometry and modulus across different noise levels. Under extreme noise, image fidelity improves by approximately 97.1% over conventional methods. We believe that this method has great potential as a robust tool for biomedical diagnosis,biomechanical characterization, and non-destructive evaluation.

0.20

HC-AINet: An Auditory-Inspired Network with Harmonic Compensation for Low-Complexity Speech Enhancement

eess.AS ·2026-10-22 ·Nan Li, Xiaoxiong Wu, Huawei Liu, Ziping Zhao
Abstract · 12 pages, 5 figures, 9 tables

Speech enhancement on resource-constrained devices requires noise suppression with limited computation and minimal loss of spectral detail. Auditory-inspired subband processing reduces computation, but its smooth gains have limited ability to distinguish harmonic peaks from neighboring noise. We propose a harmonic-compensated auditory-inspired network (HC-AINet) for low-complexity speech enhancement. A triangular equivalent rectangular bandwidth (ERB) filterbank supplies compact features to a lightweight network comprising inverted-residual units, temporal Conformer layers, and parallel frequency-gated convolutions. The network predicts global and harmonic gains together with fusion weights. A harmonic-location map derived from the noisy spectrum confines the harmonic contribution to selected bins on the original frequency grid, while the global branch controls the broader spectral envelope. The compensated magnitude, noisy phase, and a magnitude-derived minimum-phase cue jointly condition a compact phase estimator for waveform reconstruction. On VoiceBank-DEMAND, the 64-band model achieves a PESQ of 3.20 and a STOI of 0.95 with approximately 0.27 million parameters.

0.20

Learning What to Ask for Tomorrow: Active Preference Acquisition for Long-Term Personalized Web Agents

cs.CL ·2026-10-22 ·Yiqi Wang, Zizhi Zeng, Jiaqi Zhang, Yiqun Duan, Mingkai Zheng +1 more
benchmark
Abstract

Web agents increasingly serve the same user across repeated online tasks such as booking travel, shopping, and scheduling, so their long-term usefulness depends on how well they know the user's preferences. Passive memory and personalization methods cannot obtain preferences the user never reveals, and clarification methods that value a question by its benefit to the current task overlook the benefit of reusing the answer in later tasks. We study Active Preference Acquisition (APA): which missing preferences are worth the cost of asking? We define the Prospective Personalization Value (PPV) of a missing preference as its expected cumulative utility gain over future tasks minus the effort and disclosure-risk cost of asking, and select questions under a per-interaction budget. We characterize when cumulative reuse justifies acquisition despite insufficient single-task benefit, and evaluate this regime with an initial-profile diagnostic. APA computes the counterfactual gain of each candidate preference on public task prototypes with the agent's own task solver and aggregates it by the expected reuse of each task family. We evaluate it on ActiveWebPref, a controlled longitudinal benchmark with hidden but queryable user profiles and exact task utilities. At a horizon of 20 tasks, 34.4% of budget-feasible valuable missing preferences are amortization-dependent; APA acquires 80.3% of them, whereas current-task value of information acquires 5.1%. Under per-user asking-cost caps matched to APA's realized expenditure, APA achieves higher net utility than four heuristic selection baselines. In our cost setting, asking about every unknown preference yields lower net utility than not asking, while APA recovers 84.2% of its utility gain with 27.6% of its questions. The value of acquiring a user preference should account for its reuse across future tasks, not only its benefit to the task at hand.

0.20

OmniRay: Camera-Agnostic Monocular 3D Reconstruction with Composite Ray Fields

cs.CV ·2026-10-22 ·Namitha Guruprasad, Abhay Yadav, Cheng Peng, Rama Chellappa, Anand Bhattad
positioncode/data
Abstract · Project page: https://nam1410.github.io/omniray/

Monocular models recover 3D geometry from a single image in one pass, but most assume a pinhole camera and produce distorted geometry on the wide-angle imagery produced by fisheye and panoramic lenses. Camera-aware models remove this assumption by predicting a per-pixel ray field. However, they represent different camera geometries using a single fixed basis to capture both the overall camera projection and its finer variations. We argue that camera geometry can be viewed as having a hierarchy of complexity for representing rays. We present OmniRay, a feed-forward model whose Ray Module is composed of three components: a monotone radial base that captures radially symmetric structure, a spherical-harmonic residual for smooth deviations from it, and a free-form residual for remaining local variation. These components are composed as successive rotations on the sphere, allowing the representation to accommodate diverse camera geometries. Given a single image, OmniRay predicts a ray direction and an up-to-scale radial distance for every pixel, from which a 3D point cloud is recovered without needing camera information at inference time. Extensive experiments across different camera configurations demonstrate that OmniRay establishes a new state-of-the-art in camera-agnostic monocular 3D reconstruction.

0.20

FSVL: Taxonomic Fast-Slow Vision-to-Language Collaborative Framework for Interpretable Confusable Fish Species Identification and Knowledge Generation in the Middle Yangtze River

cs.CV cs.AI ·2026-10-22 ·Meng Liang, Fucheng Wang, Lang Zhao, Jun Chen, Ziying Li +6 more
Abstract

The Yangtze River Basin is a critical region for freshwater biodiversity conservation. The implementation of the Yangtze River Ten-Year Fishing Ban has increased the importance of effective aquatic biodiversity monitoring and species-level identification. However, fine-grained identification of closely related freshwater fish species remains challenging due to subtle morphological differences. Existing deep learning methods often struggle to balance classification accuracy and inference efficiency, while LLMs may generate unsupported information when producing specialized biological knowledge. To address these challenges, this study proposes FSVL, a fast-slow dual-pathway framework integrating visual recognition with LLM-based Retrieval-Augmented Generation (RAG). A lightweight ResNet50 fast path performs preliminary classification across 51 fish species in the Middle Yangtze River Basin and routes predefined groups of easily confused species to a slow path for refined identification. Guided by taxonomic criteria, the slow path uses YOLOv8n to detect key morphological regions, including the head, dorsal fin, and tail, leveraging morphometric and outline traits to improve discrimination among closely related species. The framework further integrates a retrieval-enhanced Qwen3-14B backend, using soft tokens and Low-Rank Adaptation (LoRA) for parameter-efficient domain knowledge generation. On the evaluated test set of 607 images, the fast classifier achieved 99.67% accuracy, whereas the cascade corrected the two errors, yielding 100.00% overall accuracy; further validation on independent specimens and external conditions remains needed. Across 255 knowledge generation samples, scientific name accuracy rises from 91.37% to 100.00%, genus accuracy from 90.20% to 98.82%, and ROUGE-L from 0.335 to 0.655, supporting intelligent biodiversity monitoring and fisheries management.

0.20

From Structured Records to Patient Narratives: Preserving Clinical Alignment in LLM-Generated Health Summaries

cs.CL ·2026-10-22 ·Bayram Ayadi, Florian Geissler, Johanna Schmidhuber
Abstract · 18 pages, 3 figures

As wearable and digital health systems increasingly generate longitudinal physiological records, communicating these observations in patient-accessible language remains a significant challenge. Large Language Models (LLMs) offer a promising mechanism for translating structured clinical information into readable narratives, but this process may introduce semantic drift, where clinically precise meaning is distorted during simplification. We investigate evidence-grounded health summarization for menstrual health monitoring using a constrained generation framework that combines structured evidence conditioning with Retrieval-Augmented Generation (RAG) using a prompt-based constrained generation framework. We evaluate semantic alignment using Natural Language Inference (NLI), atomic factuality metrics, and readability measures. Across three LLMs and 4,500 generated summaries, explicit grounding constraints consistently improve clinical alignment, particularly by improving uncertainty preservation and, once faithfulness is evaluated against retrieved evidence, generally preserving or improving faithfulness, while effects on omission remain model-dependent. We also find that improved clinical alignment comes at the cost of increased reading difficulty and medical terminology. These findings highlight the need to jointly evaluate faithfulness, uncertainty preservation, and readability when developing patient-facing medical LLM systems.

0.20

ResGVC: Residual Coding for High-fidelity Generative Video Compression

cs.CV ·2026-10-22 ·Ding Ding, Daowen Li, Ying Chen, Kai Li
hot direction
Abstract

Generative video compression (GVC) methods usually employ decoder-side diffusion models to denoise reconstructed signals based on generative priors. Besides, most of these methods perform denoising in latent space rather than directly in pixel domain. In this paper, we propose ResGVC, a residual coding framework for GVC that guides denoising toward original pictures by transmitting residual signals. ResGVC first encodes the original pictures within a learned video codec, and then decomposes the distortion between the codec reconstruction and the original pictures into a sequence of residual segments, where each residual segment is acquired at each denoising step. Residual segments can further be compressed by a novel residual codec with a diffusion transformer as an entropy model. Unlike prior methods, ResGVC introduces explicit source constraints throughout denoising rather than relying on decoder-side generation alone. The diffusion-based entropy model also enables more effective distribution modeling capability based on the powerful generative priors. Moreover, denoising directly in pixel space avoids the irreversible information loss caused by latent representation bottlenecks. Experimental results show that, compared with the previous state-of-the-art method, ResGVC achieves 63.7% and 83.8% BD-rate reductions in terms of LPIPS and PSNR, respectively, while also delivering superior visual quality.

0.20

Interpreting Physics in Video Encoders using Multi-Object Interaction Videos

cs.CV ·2026-10-22 ·Ayushi Mehrotra, Sai Vidyaranya Nuthalapati
Abstract

Video encoder models are increasingly used as world models for physical AI, where a policy plans by rolling out the latent, so what physics the latent holds bounds what the agent can plan over. Moreover, existing interpretability work probes quantities belonging to a single object; the quantities that make physics relational, such as momentum transfer, restitution, and friction, are defined between bodies and have not been examined. We probe relational physics attributes in two frozen self-supervised video encoders (V-JEPA2 and VideoMAE) under two-body settings and test if they transfer across environments. Our early findings are mixed. Both encoders separate elastic from inelastic collisions reliably. V-JEPA2's own predictor is less likely to predict a collision that creates kinetic energy compared to one that loses it. Momentum-violation detection appears in only one of the two encoders, a kinetic-energy readout survives a change of environment for V-JEPA2 but not for VideoMAE, and neither model holds a friction coefficient that survives a change of geometry.

0.19

Bilingual Language Models Capture Cross-Linguistic Influence in Second-Language Reading

cs.CL ·2026-10-22 ·Abderrahmane Issam, Irene E. Winther, Stefan L. Frank
Abstract · Under review

While monolingual language model (LM) surprisal is widely established as a predictor of reading times, it cannot capture the non-native (L2) reading dynamics that are shaped by the reader's first language (L1). In this work, we investigate whether the cross-lingual transfer within bilingual models aligns with human L1-driven biases during L2 reading. We evaluate bilingual models on eye-tracking data of readers from 12 diverse L1 backgrounds reading English texts. To isolate L1-specific transfer, we compare bilingual models trained on an L1 that matches the readers' L1 background to both a monolingual English baseline and bilingual models trained on other L1s. We demonstrate that matching the readers' L1 improves the prediction of L2 reading times. This predictive advantage is concentrated in late-stage eye-tracking measures, modulated by L1-L2 script overlap, and peaks among intermediate-proficiency readers. These findings provide evidence that bilingual LMs capture aspects of cross-linguistic influence, and that such influence is present in L2 reading behavior.

0.19

Confusion-Adaptive Margin Learning for Handwritten Letter Recognition: A Controlled Study on EMNIST Letters

cs.CV cs.LG ·2026-10-22 ·Azam Nouri
Abstract

Handwritten character recognition systems can achieve high accuracy while retaining concentrated errors among visually similar classes. Motivated by residual letter confusions in earlier geometry-based recognition work, we investigate whether these errors can be addressed through the training objective. We introduce Confusion-Adaptive Margin Learning (CAML), which constructs a sparse weighted confusion graph from a discovery partition held out from pilot training and applies pair-specific cosine-separation constraints during final training. CAML adds no classifier, routing rule, or inference-time parameters. We evaluate CAML on EMNIST Letters against a strictly matched cross-entropy CNN baseline, using the same architecture, final training data, deterministic augmentation, optimizer, and checkpoint selection. Eight confusion pairs are targeted: I/L, G/Q, U/V, R/V, D/O, C/E, I/J, and H/N. On the official clean test set, accuracy changes from 95.6346% to 95.6731%, and macro-F1 from 0.956326 to 0.956724. Mean accuracy across ten predefined corrupted conditions increases from approximately 95.3812% to 95.4303%; symmetric errors within targeted pairs decrease from 614 to 605 (1.47%). The global differences are small and not statistically reliable. CAML fixes 124 baseline errors but introduces 116 new errors on clean data (exact McNemar p=0.6515), and no corruption-specific difference remains significant after Holm correction. Thus, CAML does not establish global superiority, but shows that a leakage-controlled, pair-specific margin objective can alter targeted error structure with limited overall benefit for an already strong classifier.

0.19

Bezier-Guided Adaptive Flow for Cross-domain Sequential Recommendation

cs.IR ·2026-10-22 ·Ziang Lu, Yu Zhang, Xinzhe Jiang, Lei Sang, Yiwen Zhang
Abstract

Cross-domain sequential recommendation aims to alleviate data sparsity and cold-start problems by transferring useful preference knowledge across domains. However, users exhibit varying degrees of preference consistency across domains, while existing methods often rely on global alignment objectives or coarse-grained transfer strategies. Such approaches may introduce irrelevant information and provide limited flexibility in adapting semantic representations to individual user preferences. To address these challenges, we propose BGAFlow, a personalized framework that combines state-adaptive semantic evolution with user-specific transfer control. BGAFlow first establishes a shared semantic space through unified dual-domain pretraining with user profiles inferred by a large language model (LLM). Based on latent interest overlap, it then estimates a personalized horizon to determine the appropriate transfer extent for each user. To enable semantic adaptation, BGAFlow constructs a conditional Bezier curve as a geometric reference and learns a residual flow that progressively refines the trajectory based on intermediate preference states. The learned flow incorporates dynamic preference feedback to adjust transfer directions, while recommendation supervision and trajectory regularization encourage effective and smooth adaptation. Extensive experiments on real-world datasets demonstrate that BGAFlow consistently outperforms state-of-the-art baselines, validating the benefits of personalized transfer control and state-adaptive semantic evolution.

0.19

Visual Grounding First, Multimodal In-context Learning Follows

cs.CV cs.AI ·2026-10-22 ·Minhyuk Seo, Minjae Lee, Chaeeun Lee, Wei Lin, Muhammad Jehanzeb Mirza +2 more
publishedbenchmark
Abstract · NeurIPS 2026

LLMs have shown strong in-context learning (ICL) capability, but extending it to Multimodal LLMs (MLLMs) remains challenging. Prior multimodal ICL methods often rely on ICL-specific datasets for additional training or task vector extraction, which can improve performance on the same ICL benchmark used for adaptation. However, they do not generalize well to other benchmarks and often induce forgetting in previously well-solved tasks. For this reason, we identify 'visual grounding' as the key bottleneck in multimodal ICL; MLLMs often fail to attend to task-relevant visual evidence in demonstrations, instead over-relying on textual cues and producing hallucinated outputs. To address this, we propose ProCoRe to enhance visual grounding through contrastive reinforcement learning on multimodal contrastive data for improving multimodal ICL. ProCoRe generates captions for contrastive image pairs, decomposes them into propositions, and optimizes contrastive rewards that encourage alignment with corresponding images while discouraging mismatched ones. Despite never seeing ICL-formatted training examples, ProCoRe improves average multimodal ICL classification accuracy by 5.6% and captioning ROUGE-L by 13.8% over the off-the-shelf Qwen-VL-3-8B model, outperforming ICL-data-dependent state of the arts across benchmarks. Finally, we introduce PICL, a personalized multimodal ICL captioning benchmark that tests whether MLLMs genuinely understand visual demonstrations rather than relying on pre-trained semantic priors.

0.19

PnP*: A certifiable solver for Perspective-n-Point

cs.CV cs.ROmath.OC ·2026-10-22 ·Bang-Shien Chen, Caitrin E. Eaton, David M. Rosen
theory
Abstract · 28 pages, 3 figures

Estimating a camera pose from 3D-2D correspondences - Perspective-n-Point - is one of the most fundamental problems in computer vision. Existing methods either solve a minimal problem or replace the objective with an easier one. Surprisingly, no scalable certifiable solver for the actual Maximum Likelihood Estimation (MLE) exists. We propose PnP*, the first practical solver that guarantees efficient computation and returns a certifiable globally optimal solution. We prove that the underlying convex relaxation is exact under moderate noise, which to our knowledge is the first stability guarantee for any relaxation for the MLE of PnP. Furthermore, we reduce the relaxation by exploiting chordal sparsity pattern and quotient ring reduction such that the relaxation is tractable and scalable. The stability and scalability results are verified numerically, as PnP* returns a provably correct camera pose at over 10 Hz.

0.19

Query-Conditioned Liquid Temporal Grounding for Long-Form Audio

cs.SD ·2026-10-22 ·Yuelan Cheng, Yonggang Zhu, Yongqiang Chen, Peiwei Chang, Yingyi Chen +1 more
Abstract

Long-form temporal audio grounding (TAG) is challenging because target events often occupy only a small portion of a recording, while queries may impose multiple semantic constraints over a large temporal search space. We propose a query-conditioned liquid temporal grounding framework that injects query semantics through FiLM before a closed-form continuous-time (CfC) network models long-range temporal dependencies. Multiple query-relevant candidates are generated, from which a frozen large audio-language model selects the most compatible one for lightweight local boundary refinement. The pretrained backbone remains frozen, with only lightweight grounding modules optimized. Experiments on the English TAG subset of LAT-Bench show consistent improvements across MOSS-Audio and Qwen3-Omni, while ablations confirm the complementary roles of FiLM and CfC.

0.19

Beyond Best-of-$k$: Prior-Consistent DAPS for Ambiguous Nonlinear Inverse Problems

cs.CV cs.LG ·2026-10-22 ·Merve Kocabas, Benjamin A. Burns, Andreas Geiger, Kent Gauen, Sara Fridovich-Keil
Abstract

Diffusion models have become strong priors for solving inverse problems. However, nonlinear inverse problems such as phase retrieval remain challenging when multiple measurement-consistent solutions are distinguishable only by the prior; we call such measurements ambiguous. Existing diffusion-based samplers rely on local approximations during reconstruction. For an ambiguous measurement, these approximations can allow measurement guidance to favor a measurement-consistent mode that is close to a prior-supported mode but is itself not supported by the prior. Evaluation often masks this type of failure by reporting the best-of-$k$ runs, where the best is selected by comparison to the ground truth. We introduce Prior-Consistent DAPS (PC-DAPS), a variant of Decoupled Annealing Posterior Sampling (DAPS) designed to address this failure case by maintaining multiple candidate solutions within a single sampling run. As annealing noise decreases, PC-DAPS prunes and replaces weaker candidates according to their prior consistency, without any ground-truth access. PC-DAPS reduces the incidence of prior-inconsistent samples and improves sample quality on nonlinear inverse problems while preserving sample quality and diversity on linear inverse problems.

0.19

Predicting Music Valence and Arousal via Textual and Multimodal In-Context Learning in Large Language Models

cs.SD cs.AIcs.CL ·2026-10-22 ·Chi-En Dai, Chih-Pin Tan, Yi-Hsuan Yang
Abstract · Submitted to ICASSP 2027

While arousal can be reliably predicted from musical audio, emotional valence remains difficult to model due to its subjective and context-dependent nature. In this work, we investigate the potential of in-context learning (ICL) with large language models to advance continuous valence and arousal prediction. Rather than predicting emotion scores directly, our framework leverages contextual demonstrations from an auxiliary dataset to refine the initial predictions produced by a small, task-specific base model. We systematically evaluate a text-based ICL variant using lyrics and generated audio captions, alongside a multimodal variant that incorporates raw audio signals. Using the entire MER60 dataset strictly as a held-out test set, our experiments demonstrate substantial performance gains, particularly for valence. Text-based ICL improves the valence $R^2$ from 0.110 to 0.427 with 80 demonstration shots, while the multimodal variant achieves 0.346 with only 10 shots. These findings suggest that training-free ICL offers an effective inference-time strategy for current language models to predict continuous music emotion.

0.19

UserArena: Benchmarking Interactive User Simulation for Long-Horizon Shopping Agents

cs.IR ·2026-10-22 ·Jujia Zhao, Jiakai Tang, Jing Yu, Wen Chen, Jian Wu +4 more
benchmarkcommunity pick
Abstract

Agentic recommendation pursues shopper goals through sequences of context-dependent decisions. Developing and evaluating such policies requires faithful recommendation world models that capture both the interactive shopping environment and users' behavioral responses over long interactions. Existing resources rarely provide long-horizon, fine-grained real-user trajectories or support trajectory-level evaluation and offline validation against production A/B results. We introduce UserArena, a real-world e-commerce interactive benchmark containing 64,729 session trajectories from 11,010 users, spanning 566,102 items and 13 action types. UserArena pairs these trajectories with an interactive environment and supports offline validation against outcomes from a controlled production A/B test, enabling five-level evaluation from output validity and local decisions to sequential behavior, personalization, and outcomes. To improve simulator fidelity over these long-horizon interactions, we further propose On-Policy Trajectory Distillation (OPTD), which uses trajectory-level intent to provide dense teacher targets on student-generated rollouts. Experiments show that OPTD achieves the strongest overall fidelity among the evaluated simulators under UserArena's evaluation protocol and most closely reproduces the production policy effect in offline simulation.

0.18

DRAMA-Eval: A Benchmark for Context-Aware, Multi-Turn, Multi-Speaker Speech Generation

eess.AS cs.SD ·2026-10-22 ·Jingbin Hu, Zihang Yang, Xiaoge Peng, Haoyu Zhang, Ziyu Zhang +19 more
benchmarkcode/data
Abstract

Recent advances in speech generation are shifting toward context-aware, multi-turn, and multi-speaker synthesis. While these capabilities are in high demand for applications such as audiobook and audio drama production, evaluation has not kept pace with these advances. Beyond traditional zero-shot Text-to-Speech (TTS), modern systems in this domain primarily fall into three categories: instruction-controllable generation models, multi-turn dialogue models, and the audio branches of audio-visual generation models. However, existing benchmarks evaluate these paradigms in isolation, failing to provide a unified assessment of context modeling, instruction following, and multi-speaker generation. Meanwhile, recent progress in audio understanding models has made it feasible to evaluate such complex scenarios using LLM-as-a-Judge approaches. To bridge this gap, we introduce DRAMA-Eval, a comprehensive benchmark for evaluating such systems. DRAMA-Eval utilizes structured script contexts and offers a holistic evaluation framework covering objective, local, and global aspects of generation quality, enabling the assessment of acoustic fidelity, sentence-level execution, and long-range coherence. We benchmark leading open-source and commercial models on a human?verified dataset. The results reveal critical weaknesses in long-range speaker consistency and implicit instruction following that require deep contextual understanding. To facilitate reproducible research, we release a 12-hour bilingual (Chinese-English) test set of manually corrected real-world scripts, accompanied by ground-truth audio, evaluation code, and benchmark results.

0.18

A Model-Agnostic Correction Framework for Short-Term Building Load Forecasting

cs.IR cs.LG ·2026-10-22 ·Shiqiao Zhou, Huanbo Lyu, Holger Schöner, Zipeng Wu, Jessica Deuschel +3 more
Abstract · Work in progress

Short-term building load forecasting remains challenging because load patterns can change rapidly with weather and operating conditions. Data-driven forecasting models may become less reliable when the current operating regime is rare or poorly represented during training, even when similar situations exist elsewhere in the historical record. Recent advances in retrieval-augmented time series forecasting offer a promising way to adapt predictions to changing conditions by directly reusing relevant historical cases at inference time. Motivated by this idea, we propose WeL-RAC, a model-agnostic Weather-Load Retrieval-Augmented Correction framework that explicitly retrieves such evidence at prediction time. WeL-RAC retrieves historical cases that are similar to the current condition based on both recent load states and horizon-aligned weather forecasts. Their observed future load trajectories are then aggregated and used to correct the initial prediction, without retraining or modifying the base forecaster. We evaluate WeL-RAC on 31 building cases across three datasets, eight forecasting backbones, and four horizons from 1 to 18 hours. The framework improves the corresponding base forecasters in 90 of 96 dataset-level comparisons. Beyond accuracy, success and failure case studies further illustrate the framework's interpretability by tracing how retrieved historical evidence improves or degrades forecasts. These findings support explicit historical retrieval as an effective and interpretable complement to the building load forecasting.

0.18

Through a PRISM, Clearly: Step-Sequence Reasoning for Mistake Recognition in Procedural Videos

cs.CV cs.CL ·2026-10-22 ·Olga Loginova, Oleksandr Bezrukov, Frank Keller
benchmark
Abstract

Procedural mistake recognition requires identifying deviations from valid multi-step activities. We study post-completion recognition from videos with benchmark-provided step boundaries. We introduce PRISM (Procedural mistake Recognition via Interpretable Semantic Modeling), which grounds video segments in language and maps their descriptions to procedural states using a lightweight hidden Markov model trained on correct executions. Reference alignment and semantic checks identify errors in step content and order. To support evaluation across heterogeneous benchmarks, we map dataset-specific annotations to four edit-based error types: insertion, deletion, substitution, and transposition. Across three benchmarks, we compare gold and VLM-generated descriptions in raw, semantic-role, and action-object forms. Decoder controls show consistent gains from the trained local decoder over nearest-candidate matching, with additional, dataset-dependent benefits from sequence decoding. Semantic role labeling improves Any-F1 in all six matched grounded comparisons, although the gains vary across datasets and do not translate into uniform improvements under separate configuration selection. These results establish language grounding as an effective interface for explicit procedural mistake reasoning and characterize the contributions of representation and decoding.

0.18

A corpus-only AUC diagnostic for pooled speech-based depression detection

eess.AS ·2026-10-22 ·Cezary Polak, Milosz Dudek, Daria Hemmerling
Abstract

When clinical speech corpora are pooled for evaluation, corpus membership can itself become predictive if class prevalences differ across sources. We derive the AUC of an oracle predictor that uses only corpus identity and empirical prevalence, a diagnostic computable directly from corpus composition. Across five depression corpora in three languages (590 speakers), the corpus-only AUC is 0.697, and corpus identity is predicted from six acoustic and self-supervised representations with 97.5-99.8% accuracy. Permuting depression labels within each corpus removes label structure while preserving prevalences; the residual pooled-minus-LOCO discrepancy then increases across three nested configurations in the same order as the diagnostic. Pooled 5-fold cross-validation also outscores corpus-held-out evaluation across all six representations. Prevalence equalisation substantially reduces the discrepancy for eGeMAPS and WavLM-Base, whereas a sample-size-matched control does not. Corpus-specific base rates therefore create a measurable shortcut opportunity in pooled clinical-speech evaluation, motivating corpus-aware diagnostics alongside pooled performance.

0.18

Lightweight Frame Compensation for Real-Time Polyp Segmentation under Frame Skipping

cs.CV ·2026-10-22 ·Weihao Gao
Abstract · Subsequent analysis revealed a flaw in the evaluation protocol of the proposed lightweight frame compensation. The original experiments did …

Real-time polyp segmentation in computer-assisted colonoscopy is hampered by the difficulty of deploying high-accuracy models under strict latency budgets. Frame skipping, which performs inference every K frames and reuses or interpolates the rest, improves throughput but introduces motion blur and lesion misalignment that degrade temporal consistency. To close this gap, we propose the Lightweight Frame Compensator (LFC), a 77K-parameter plug-and-play module inserted before a frozen segmentation backbone. Operating on a three-frame window, LFC fuses consecutive frames via spatio-temporal attention into a single compensated frame, and is trained only on pseudo-labels generated by a differently seeded model, without any expert annotation. Experiments on three polyp video benchmarks show that LFC recovers most of the accuracy degraded by direct frame skipping and, in several settings, reaches the accuracy of frame-by-frame inference, while retaining a 2.05X--2.55X throughput gain. We further evaluate the temporal stability of three backbones spanning CNN, pyramid Transformer, and global-attention Transformer architectures, revealing a spectrum of sensitivity to frame skipping that arises from how locally or globally each architecture aggregates misaligned input. The compensation gain of LFC follows this spectrum: it is largest for high-capacity Transformer backbones, which achieve the best static accuracy but suffer the most under skip-frame inference. LFC thus offers a lightweight, pluggable, and deployment-friendly route to real-time polyp segmentation.

0.18

Same Entity, Different Script: Diagnosing Cross-Script Entity Matching Failures in Language Models

cs.CL cs.AI ·2026-10-22 ·Athira Gopal, Patil Prathamesh Pravin, Ankush Kadu, Arpit Jain, Ashwanth Krishnan
Abstract

Multilingual language models can give different identity judgments when the same people's names are written in different scripts. We study this effect using 601 audited identity pairs with attested Devanagari and Latin forms, holding the question and relation fixed. Qwen2.5-7B's same-person accuracy falls from 35.1% in Latin to 8.5% in Devanagari under Marathi prompts, while different-person accuracy remains above 98%. Evaluation across five models reveals distinct response biases, and reduced tokenization fragmentation does not reliably close the gap. Replicating the design on 374 audited Malayalam pairs built from the same entities reproduces the effect in a second script and language family: same-person accuracy falls to 5.3%, about half the matched Devanagari rate, while the margin still ranks matching above non-matching names at 0.925 AUROC. Latin-trained probes recover label information from Devanagari states, exceeding a string-feature baseline under similarity-matched negatives. Scalar calibration recovers 87% of the raw model-probe balanced-accuracy gap, identifying decision criterion placement as a major contributor. Activation patching localises causal sensitivity to the second entity in Qwen's middle layers; one shared sensitivity direction transfers about half the full-state margin correction. A fact-independent intervention at this site raises held-out Marathi same-person accuracy from 6.5% to 33.5%, with 98.5% different-person accuracy. Together, these results distinguish script-dependent discrimination from criterion placement and identify a causal site for improving identity judgments.

0.17

RECAP: Looking Once Is Not Enough for Vision-Language Reasoning

cs.CV ·2026-10-22 ·Zhaolu Kang, Tailong Luo, Chenxin Li, Yingjie He, Zhenyu Yu +11 more
published
Abstract · NeurIPS 2026

Long chain-of-thought reasoning improves vision-language models (VLMs), but it also exposes a temporal grounding failure: as generation unfolds, models may drift from image evidence while reinforcement learning still provides only a final-answer reward. We study this gap through visual revisit, a temporally localized reactivation of image evidence during reasoning. Across three VLM backbones, revisit peaks are predictive of correctness and causally linked to performance: masking them degrades accuracy more than masking non-peaks, while injecting revisit-like peaks into failed rollouts partially restores correct answers. Control analyses show that this signal is not explained by raw attention strength, uncertainty, logit margin, or hidden-state norm. We propose RECAP, a GRPO-compatible training method that turns detached revisit traces into a credit-assignment signal. RECAP uses revisit-conditioned advantage estimation, dynamic discounting, and gated value prediction to propagate final rewards through visually important reasoning steps, without extra rewards, additional rollouts, or inference-time changes. Across Qwen2.5-VL-7B, InternVL3-8B, and LLaVA-OV-7B, RECAP consistently improves over GRPO and VPPO, yielding $+6.1$--$+6.5$ HallusionBench gains over the SFT base while preserving reasoning and text-only performance. Results on larger LoRA-tuned models, Chinese zero-shot benchmarks, visual perturbations, and human evaluation further demonstrate robust improvements in image-grounded reasoning.

0.17

PAIR: Preference-Aligned Intent Retrieval for Premium People Recommendation at LinkedIn

cs.IR cs.LG ·2026-10-22 ·Zhongkai Shangguan, Yanbin Kang, Yihan Cao, Ruijie Jiang, Qiannan Yin +5 more
Abstract

General-purpose embeddings trained on profile similarity often misrank candidates in professional networks, where relevance is inherently complementary rather than symmetric. For example, job seekers usually engage with recruiters rather than peers with similar credentials. Current user recommendation systems are often built on profile similarity or social signals such as shared connections, which are effective in friend-oriented social networks where users tend to connect with similar people or within existing social circles. However, this does not hold in professional platforms like LinkedIn, where interactions are often goal-driven and cross-role interactions, requiring matching based on complementary intents rather than similarity or social proximity. In this paper, we propose Preference-Aligned Intent Retrieval, a retrieval framework that is designed to model user intent preference instead of profile similarity. Our approach is inspired by the preference-learning perspective of Direct Preference Optimization, but instantiates preference learning as an exposure-conditioned multi-negative contrastive objective over embeddings. This formulation simplifies training and scales efficiently to large candidate sets without requiring an explicit reference model. Furthermore, we incorporate user interaction history by aggregating click histories with positional encoding to capture temporal order, which enables constant-cost updates of user representations over a fixed history window, allowing new interactions to be incorporated without re-encoding history. As a result, personalization can be updated efficiently using lightweight CPU operations without repeated LLM inference during serving. We evaluate PAIR on LinkedIn's Premium retrieval task. A 0.6B parameter model achieves a MAR@100 of 0.45.

0.17

ICGF-CXRMate: Image-Conditioned Clinical Context Compression and Gated Fusion for Chest X-ray Report Generation

cs.CV ·2026-10-22 ·Vu Minh Tran, Thuc Huynh, Son Thanh Luu, Doanh C. Bui
Abstract

Auxiliary patient data can improve chest X-ray report generation, but directly concatenating heterogeneous clinical information with visual tokens produces long, variable-length decoder prompts and may expose the model to redundant or image-irrelevant context. We propose ICGF-CXRMate, an image-conditioned clinical context compression and gated fusion framework. Learnable queries are first conditioned on the current chest X-ray representations and then attend to the patient-data sequence, compressing it into a fixed number of clinical context tokens. Gated cross-attention selectively integrates these tokens into the visual stream, while the compressed tokens are retained in the decoder prompt as a direct clinical pathway. Experiments on linked MIMIC-CXR and MIMIC-IV-ED examinations show that ICGF-CXRMate improves report quality while reducing the average prompt length. In Stage~2, it improves RadGraph-F1 from 24.00 to 28.86, CXR-BERT similarity from 56.86 to 62.80, and BLEU-4 from 5.14 to 7.96 compared with CXRMate-ED, while reducing the average prompt length from 373.89 to 288.37 tokens, corresponding to a 22.9\% reduction. After self-critical sequence training, ICGF-CXRMate outperforms CXRMate-ED on six of eight report-quality metrics in Stage~3. Ablation studies confirm the complementary contributions of image-conditioned compression, gated fusion, and direct retention of the compressed clinical tokens.

0.17

WildDepth-v2: A Synchronized RGB--LiDAR--Radar Dataset and Benchmark for Metric Depth, Motion, and Tracking of Wildlife

cs.CV ·2026-10-22 ·Muhammad Aamir, Sangyun Shin, Naoya Muramatsu, Matthew Wijers, Amir Patel +2 more
benchmark
Abstract

Measuring animal distance and motion in metric units underpins abundance estimation, morphometrics, and movement ecology, yet monocular video lacks metric scale and wildlife datasets lack metric ground truth. We present WildDepth 2, extending the synchronized RGB-LiDAR WildDepth dataset in three directions. (i) Imaging radar: 37 RGB-LiDAR-4D-radar sequences (19,716 frames) with per-point Doppler velocity and a measured error budget, to our knowledge the first imaging-radar wildlife recordings. (ii) Benchmarking: a frozen, checksummed hidden test set for RGB-only distance estimation and RGB+sparse-LiDAR depth completion, evaluated separately on animal and background pixels. (iii) Motion and tracking: metric 3D trajectories obtained by lifting 2D tracks using LiDAR, radar, or monocular depth. Across 48,723 animal frames, object-level radar calibration improves monocular depth over scene-level LiDAR calibration without requiring LiDAR at inference (AbsRel 0.120 vs. 0.198), although animal-level LiDAR remains more accurate (0.052). Across 431 tracks, once-per-track radar scaling reduces 3D position error four-fold (2.20 to 0.58 m) and distance-sampling detection-function bias from -13% to -2%, matching once-per-track LiDAR calibration. Transferring radar Doppler onto LiDAR produces motion masks 3.7 times denser than radar alone at 72% precision, reduces moving-animal LiDAR accumulation error by 37%, and estimates radial speed (r = 0.60), albeit with approximately 50% underestimation.

0.17

STEP: Surgical Point Tracking from Sparse Endpoints and Pseudo-trajectories

cs.CV ·2026-10-22 ·Bing Fan, Anwesa Choudhuri, Zhongpai Gao, Meng Zheng, Benjamin Planche +3 more
Abstract

Surgical point tracking requires persistent correspondence through tissue deformation and occlusion, yet dense trajectory annotations are costly to obtain. We present STEP, a two-stage framework that learns from surgical video and sparse physical endpoints. First, a frozen offline teacher generates candidate trajectories, and temporal reversibility and stereo consistency select dense supervision for a causal student. Second, a robust physical endpoint loss anchors terminal correspondence, while a round-trip loss constrains temporal consistency. A gradient-free prefix exposes training to accumulated drift, while a final differentiable window bounds activation memory. The trained tracker processes a single RGB stream with fixed weights. In STIR-to-VL-SurgPT transfer, temporal-stereo union selection reduces endpoint error by 7.0% relative to unfiltered trajectories. On a post-hoc diagnostic of 29 VL-SurgPT clips excluded from the endpoint-training manifest, STEP reduces endpoint error from 31.52 to 16.80 pixels relative to dense-label fine-tuning. These results support combining curated trajectories with physical observations to train surgical point trackers.

0.17

AudioLift: Audio-Driven Single Image to Full-Head Gaussian Avatar via Multi-view Video Generation

cs.CV ·2026-10-22 ·Zhuoyuan Jiang, Weijie Lyu, Xiaoyuan Wang, Tiancheng Shen, Ming-Hsuan Yang
code/data
Abstract · Project page: https://zhuoyuanjiang.github.io/audiolift-page/

Existing 3D talking avatar methods mainly rely on parametric face models including 3DMM or FLAME, whose compact motion representations limit their ability to synthesize extreme expressions and to model accessories such as headwear. We introduce AudioLift, a two-stage framework that uses generated multi-view videos as the intermediate representation for audio-driven avatar generation without parametric face models or per-instance optimization. Our framework combines a synchronized multi-view video generation stage with an inconsistency-robust Gaussian reconstruction stage to recover coherent full-head geometry from a single portrait image. Extensive experiments demonstrate that AudioLift produces 3D talking avatars capable of capturing more extreme facial expressions than existing methods, with improved geometric completeness and strong multi-view consistency across the full $360^\circ$ viewing range.

0.16

LoRA Superposition across Pretrained Models

cs.CV cs.AI ·2026-10-22 ·Keigo Watanabe, Hiroshi Kera, Kazuhiko Kawamoto
Abstract · 12 pages, 5 figures, 2 tables

Low-rank adaptation (LoRA) is usually tied to the pretrained backbone on which it is learned. We ask whether LoRAs trained independently for the same task on different pretrained backbones can be summed and applied unchanged to all contributing backbones. We call this construction LoRA superposition. Across seven ViT-B/16 CLIP backbones and eight image-classification tasks, the exact sum improves all 56 backbone-task pairs and achieves a mean accuracy gain of 30.21 percentage points over the unadapted backbone. The exact sum achieves a mean accuracy only 0.28 percentage points below that of individually trained solo LoRAs. In contrast, a leave-one-model-out (LOMO) sum produces a mean gain of only 0.06 percentage points over the unadapted backbones. Backbone-specific solo LoRAs are nearly orthogonal in parameter space, while pairwise cross-backbone responses are strongly concentrated on the diagonal. These results support a component-wise non-interference hypothesis, under which the frozen backbone acts as an implicit selector of its own component without explicit routing or cross-model alignment.

0.16

Signed Wasserstein distance for room impulse response comparison

eess.AS ·2026-10-22 ·Linda Fabiani, Sebastian J. Schlecht, Filip Elvander
Abstract

The nature of room impulse response signals is often adverse to a clear evaluation of similarity: they are sparse, rapidly varying, signed signals resulting from the complex interaction of room reflections and phase relationships. Therefore, conventional point-wise metrics may be sensitive to small temporal shifts and provide poor loss surfaces for optimization. In this paper, we propose an optimal transport-based distance for RIR similarity that explicitly accounts for the structure of impulse responses, inspired by the signed Wasserstein distance formulation. Optimal transport provides a natural framework for this problem, by comparing signals based on how their underlying distributions are displaced rather than requiring a one-to-one correspondence. We evaluate the proposed distance against commonly used audio baseline metrics through a series of room-parameter perturbation studies. The results demonstrate the advantage of using the signed Wasserstein metric for geometric evaluation tasks and show a promising performance for room parameter optimization within a differentiable Image-Source model.

0.16

The Delayed Countdown Method for Validating Review-Oriented AI Workflows in eDiscovery

cs.IR ·2026-10-22 ·David D. Lewis, Tara Emory, Lenora Gray, Mark Noel, Jeremy Pickens +1 more
Abstract

Review-oriented workflows have become the dominant approach to AI-assisted document review in electronic discovery (eDiscovery). How to validate the effectiveness of review-oriented workflows has, however, remained a confusing and controversial topic in technology-assisted review (TAR) in the law. We examine the sources of this controversy, and propose a simple and practical evaluation approach for review-oriented workflows. Our Delayed Countdown method is easily incorporated into existing review-oriented workflows and software, is compatible with arbitrary stopping rules, and produces valid frequentist confidence intervals on effectiveness. We explain the conceptual foundations of the method, provide example computations, discuss the handling of logically impossible recall values, and suggest future work on decreasing review costs. (This paper was submitted October 1, 2026 to The Tenth Annual Workshop on Applications of Artificial Intelligence in the Legal Industry.)

0.16

RecallGS: Geometry-Grounded Recall for Online Semantic Gaussian Splatting

cs.CV cs.AI ·2026-10-22 ·Thanh-Khoi Nguyen, Thien-Phuc Tran, Minh-Triet Tran
Abstract

Online semantic 3D Gaussian Splatting (3DGS) enables open-vocabulary scene understanding from streaming observations. However, existing methods recover geometric context missing from image-derived semantic features using an additionally learned 3D network, incurring extra training and runtime overhead. We observe that such geometric-semantic evidence may already exist in the agent's observation history and can be retrieved rather than learned. Based on this insight, we introduce RecallGS (Geometry-Grounded Recall for Online Semantic Gaussian Splatting), a lightweight, training-free mechanism that treats past observations as a non-parametric geometric-semantic prior. RecallGS follows a recall-ground-fuse pipeline: CLIP retrieves relevant observations, alpha-compositing weights identify geometrically supported views, and aggregate their semantic features over each Gaussian's rendered footprint. On ScanNet and ScanNet++, RecallGS achieves strong semantic segmentation performance at 2.48 FPS, demonstrating that geometric-semantic context can be effectively recovered from observation history without an additional learned 3D branch.

0.15

Learning from Next-Frame Prediction: Towards Unified and Scalable Visual World Modeling

cs.CV cs.AI ·2026-10-22 ·Yifeng Yao, Jinghan Li, Zhihao Zhang, Yajiao Xiong, Yang Jin +3 more
Abstract

Video provides a natural source of supervision: what happens next serves as a learning signal for visual world modeling. Yet masked video pretraining focuses on recovering missing content from surrounding context rather than predicting the future from the past. Autoregressive methods instead predict forward one token at a time, but have yet to match joint-embedding methods in representation quality. This gap motivates us to revisit the granularity and space of future prediction. In this work, we introduce VANE, a self-supervised visual pretraining framework built on next-frame prediction in embedding space. A causal encoder summarizes past observations for a state predictor that infers the representation of the entire next frame, making temporal evolution the primary learning signal. This prediction is carried out in a fixed embedding space provided by a frozen semantic tokenizer, emphasizing scene content and structure. To retain fine-grained spatial detail alongside semantic content, a lightweight auxiliary generative decoder reconstructs future-frame patches conditioned on the predicted representations. To unify visual modeling across images and videos, we construct a local-to-global crop sequence for still images, allowing the same encoder to predict newly revealed regions from earlier crops under the same objective. Experiments show that VANE improves with model and data scaling, achieving SOTA on most anticipation, recognition, and dense perception benchmarks among unified encoders. Especially on EPIC-KITCHENS-100 action anticipation, it surpasses V-JEPA 2.1 by 7.1 points in mean-class recall@5. These results demonstrate that next-frame prediction in embedding space is effective and scalable for visual world modeling.

0.15

OpenHome: From Perception to Insight in Smart-Home Camera Videos

cs.CV cs.AI ·2026-10-22 ·Xianglong Wang, Neil Rodrigues, Pei Guo, Yiru Fang, Kevin Beussman +7 more
benchmark
Abstract

Vision language models (VLMs) have been increasingly adopted in consumer smart-home camera systems to interpret recorded videos and provide insights. However, video clips captured from smart-home cameras are different from curated web videos dominating existing VLM benchmarks. Smart-home video clips often do not contain subtitles or narration and are captured from different viewing angles, under various lighting conditions, and in diverse scenarios. However, the capabilities of the model in this domain, under sparse visual signals and limited context, are not fully assessed by existing VLM benchmarks. To close this gap, we introduce OpenHome, a visual-only, four-option multiple-choice benchmark for evaluating smart-home video understanding. It contains 2,254 questions on 1,118 real-world video clips and organizes 10 tasks into three capability levels. Perception evaluates visible content, Reasoning evaluates causal, spatial, and temporal relationships, and Insight evaluates subject intent, supporting evidence and appropriate homeowner responses. Based on the dataset, we describe the task design, data pipeline and evaluation protocol, and report results for 32 closed-source and open-weight VLMs. The best model, Gemini 4 Argon, reaches 81.2% overall accuracy. Frontier closed-source models outperform open-weight models on every level. The best closed-source model leads the best open-weight model by 16.3% on Perception, 22.4% on Reasoning and 37.4% on Insight.

0.14

Interpretive Variation and Discourse Structure in Detecting Enthymemes in Controversial Political Discourse

cs.CL ·2026-10-22 ·Martial Pastor, Nelleke Oostdijk
Abstract

Enthymemes are arguments that leave a premise or conclusion unstated, and they are a common persuasion device on social media. Because readers must fill in the missing part themselves, they often disagree on whether a text contains one. This paper asks how to detect and evaluate enthymemes when such disagreement is part of the task. On tweets about vaccination and immigration, each labeled by five annotators, F1 stalls around 0.67 for every fine-tuned model we train, regardless of data size, added features or label type. We therefore ask how these models should be evaluated instead, and propose metrics that compare predictions with the full distribution of annotator votes, together with a model that predicts this distribution directly and captures disagreement better at no cost in F1. We then ask what role discourse structure plays. Models learn to rely on explicit argumentative structure: they reliably recognize tweets without an enthymeme but rarely commit on tweets that contain one. Both the models and a discourse parser miss arguments whose links are left implicit and must be inferred by the reader. Finally, we discuss what these blind spots mean for content moderation systems trained on majority labels.

0.14

Lip-Focused Audio-Visual Consistency Detects Speech Content Manipulation beyond Coarse Synchronization

eess.AS ·2026-10-22 ·Jian Zong, Chenghan Lin, Meng Liu, Zhao Xu, Junyu Wang +4 more
publishedbenchmark
Abstract · Accepted by NCMMSC 2026, LNCS format, 2 figures, 5 tables

Speech-content manipulation changes what a visible person appears to say by replacing the speech track and, in stronger attacks, regenerating mouth motion to match the forged utterance. Such attacks are difficult to expose with texture artifacts or coarse lip-sync scores because modern retalking models explicitly optimize audio-visual synchrony. We test the narrower hypothesis that speech replacement still perturbs the local relationship between acoustic dynamics and mouth motion after speaker identity or coarse synchrony is controlled. To isolate this signal, we introduce SyncSpeech, an LRS3-derived benchmark containing real videos, audio-splicing cheapfakes and Wav2Lip retalking attacks under same-speaker and different-speaker replacement. We then introduce C-MIND, a lip-focused detector that aligns audio and mouth-motion representations at frame level and models their agreement and disagreement without transcripts, automatic speech recognition features or an unmanipulated reference. On SyncSpeech, pretrained SyncLip-C achieves 95.813% scenario-averaged accuracy and 0.9943 AUC, outperforming sentence-level fusion, full-face inputs and a reproduced SyncNet baseline under the reported protocol. When trained on the corresponding FakeAVCeleb training split, C-MIND correctly classifies 278 of 280 samples in the selected unseen-speaker test subset, whereas accuracy falls to 82.775% on unseen Video-Retalking attacks. These results identify lip-focused speech-articulation consistency as a useful but generator-sensitive forensic signal within the evaluated settings.

0.14

KinoUAV: A Benchmark for Safe and Stable UAV Vision-Language Navigation with Kinodynamic Modeling

cs.CV ·2026-10-22 ·Chenyao Yang, Peijin Wang, Yingchao Feng, Wenhui Diao
benchmark
Abstract

Unmanned aerial vehicles (UAVs) have recently emerged as important remote sensing platforms, and UAV navigation has shown great potential in remote sensing applications. However, existing methods do not explicitly incorporate kinodynamic modeling or safety distances into trajectory generation, resulting in high absolute yaw rates, frequent collisions, and low navigation success rates, which limit safety and motion stability. To address this challenge, we introduce KinoUAV, a kinodynamic benchmark comprising a dataset and KinoUAV-Agent. First, we develop a hierarchical kinodynamic trajectory generation pipeline that drives UAV through acceleration actions. It uses waypoint chains from global planning and generates trajectories segment by segment under velocity, acceleration, and jerk constraints, reducing oscillations and improving trajectory stability. Furthermore, we introduce directional free-space distance (DFD) labels into the constructed dataset to describe the available clearance for safe motion decisions. The proposed KinoUAV-Agent predicts DFDs and uses them to generate safe actions, decoupling perception from action decision-making. This design enables autonomous obstacle avoidance and improves navigation safety. The benchmark incorporates the proposed Cruise-Seek task, in which the UAV first enters the target region and then searches for the target, reducing the difficulty of target search and improving navigation success. Experiments on KinoUAV show that KinoUAV-Agent outperforms existing methods, achieving the highest success rate while maintaining lower absolute yaw rates and collision rates. These results demonstrate the effectiveness of kinodynamic modeling for safe and stable UAV navigation.

0.14

MixJudge: Learning a Language-Queryable Critic for Vocal Mixing

cs.SD cs.CL ·2026-10-22 ·Kuan-Hsun Ho, Yuan-Chiao Cheng, Yen-Tung Yeh, Yi-Hsuan Yang
Abstract · Submitted to ICASSP 2027

Mixing requires not only applying processing but also assessing the resulting mix. Audio-language models can provide conversa- tional mixing advice, but existing evaluations do not separately test whether they can judge specific mixing issues. Since professional supervision for such judgments is scarce, we construct a curriculum of 11 vocal-mix issues at multiple severity levels and use it to train MixJudge, a language-queryable audio critic. MixJudge shows its ability to retrieve mixing issues, distinguish opposite issues, and order the severity of an issue. Moreover, its judgments transfer to menu selection: given a set of processing candidates, MixJudge often favors versions acoustically closer to a held-out professional version, demonstrating how explicit mixing judgment can support comparison and downstream mixing decisions.

0.14

Retrieval-Augmented Online Textual Memory for Episodic Memory Video Question Answering

cs.CV ·2026-10-22 ·Raffaele Cali, Giuseppe Lando, Rosario Forte, Antonino Furnari
Abstract

Long egocentric videos pose a central challenge for multimodal language models: the visual stream must be compressed into a memory representation that can be searched efficiently while still preserving the evidence needed for question answering. Building on an online textual-memory framework in which first-person videos are converted into temporally grounded captions, this work systematically studies the role of retrieval in episodic-memory video question answering. We evaluate how different retrieval units, query formulations, context budgets, and reranking strategies affect multiple-choice answering on QAEgo4D-Closed, isolating retrieval as the main variable while keeping the underlying generated memory fixed. Our best configuration, Parent, matches fine-grained caption subchunks, then provides their full parent captions to the reasoner. It achieves 56.4 % accuracy, matching strong full-memory and retrieval baselines while exposing only a small fraction of the textual memory at question time. More importantly, our analysis shows that better temporal coverage does not necessarily translate into better answers: increasing the retrieval budget improves temporal Hit@K and Recall@K, but can reduce accuracy by introducing visually plausible distractors and uncertain caption evidence. These findings suggest that retrieval-augmented egocentric memory should not be optimized only for evidence recall, but also for the reliability, specificity, and answer-discriminative value of the retrieved context. The study motivates more structured episodic memories that move beyond raw caption retrieval toward evidence-aware representations for long-form egocentric reasoning.

0.14

WorldForge V2: Eliciting Camera-Trajectory Alignment in Video Diffusion Models

cs.CV ·2026-10-22 ·Chenxi Song, Yanming Yang, Tong Zhao, Ruibo Li, Joey Tianyi Zhou +1 more
Abstract

Video diffusion models have learned rich 3D/4D priors, making them promising for scene generation and video re-rendering. However, their use in spatial tasks is limited by imprecise camera control, spatial and temporal inconsistencies, and entangled camera motion with scene dynamics. Existing approaches, such as task-specific fine-tuning and post-process warping, often introduce visual artifacts, generalize poorly, or incur high computational costs. We propose WorldForge V2 to achieve accurate camera control while preserving the generative capabilities of pretrained video models. We first develop a training-free guidance framework using warped references to iteratively refine generation along a target camera path. Optical-flow analysis identifies motion-related latent channels for selective guidance, helping separate motion control from appearance. A dual-path strategy further compares guided and unguided predictions to correct artifacts from imperfect references. Together, these components guide viewpoint changes while preserving scene appearance and object motion. To reduce refinement costs, v-cache approximates the correction using the preceding denoising prediction, avoiding an additional model evaluation. We further extend motion localization from latent channels to network blocks and introduce Trajectory LoRA, which concentrates adaptation on blocks most responsive to trajectory guidance. A two-stage training strategy teaches these adapters to denoise the states encountered during guided generation, using limited supervision effectively. This enables reliable camera control with very little training data, with a single training scene already supporting generalization to unseen scenes. Quantitative and qualitative evaluations across multiple datasets show that our method achieves state-of-the-art generation quality and trajectory accuracy in camera-controlled video generation.

0.14

Ask About the Right Moment: Training-Free Object Identity Reasoning for Video-Language Models

cs.CV ·2026-10-22 ·Di Yang, Xiaochang Li, Muyang Liu, Yufei Wang, Yanhai Xiong
Abstract

Many questions about a video concern one physical object. Where does the ball end up after the cups are shuffled? Which card is the queen? How often was the mug covered? Open video-language models (Video-LMs) recognise such objects in single frames but do not keep them apart over time. On shuffled shell games, open models from eight families stay near chance, and so do published tool agents. We separate semantic identity binding from temporal identity maintenance. Binding should happen at a moment when the evidence for the object is visible, and maintenance should be left to a tracker. Track, Then Bind implements this principle without training. A promptable video tracker follows every candidate object and draws it with a track number. A frozen vision-language model (VLM) answers one question about the identifying evidence, which binds the target to a track. The answer is then derived from a text description of the tracks. With Qwen3-VL-32B, the agent solves both VET-Bench tasks, where a fine-tuned trajectory model reaches 92% and 90%. On real cup games from the Perception Test, it raises accuracy from 35.5% to 52.3%. Without retuning, the same configuration also solves VET-Bench with Gemma 4 and GPT-5.5. Its track-derived visibility timelines add 8.8-17.9 points on TOC-Bench for open models. GPT-6.1 already answers 92.5% of the real cup games directly but 71.0% through our pipeline, while our timelines still raise its TOC-Bench accuracy from 52.5% to 57.7%. The benefit of external tracking thus depends on the underlying model and the task.

0.13

From Clinical Judgment to Checkable Rules: Clinician-Led Guardrails for Mental-Health Chatbots

cs.CL cs.AIcs.CY ·2026-10-22 ·Chenhan Lyu, Kit Wislocki, Sarah Elizabeth Stoeckl, Maitreyi Sinha, Amir Rahmani
benchmark
Abstract

Large language models (LLMs) deployed in mental-health contexts can be evaluated for safety failures, but the resulting judgments are often difficult to inspect or explain. Common approaches to improving safety rely on prompt engineering, LLM-as-a-Judge, and Constitutional AI. In this work, we describe a clinician-led alternative in which two clinicians define, finalize, and adjudicate the four safety checks of a rule base, which is formally verified for internal consistency and applied as a guardrail over LLM outputs. Here and in the title, clinician-led refers to this rule base, not to the fact extractor or the uncertainty default, which are ours. Every verdict of the resulting verification layer is recomputable from a named rule and sixteen facts. Against an LLM judge matched in model, revision, quantization, rule text, and decoding but without the layer's computed 988/911 fact, the largest disagreement class, the satisfied-rule class, is structured: the Llama-3.3-70B judge, on its primary prompt, demands a crisis resource from 55 of the 65 replies (84.6%) that already contain one, while the same model, asked as a fact extractor, reports a recognized resource present on 54 of the 55. The verification layer never raises that demand, by construction. We report with equal prominence where this judge is stronger: fewer false crisis alarms on the one reference-scored check. We also report the measured cost (a specification revision raised block-class outcomes on held-out boundary prompts of an over-refusal benchmark from 6.0% to 16.0%, chiefly by changing fact extraction), the clinicians' decision to keep our uncertainty default, and the analysis of the result. The contribution is not a safer verdict. It is a guardrail whose disagreements with the judge localize to named rules and facts and whose measured costs trace to named facts and the uncertainty default.

0.13

Asking Less, Adapting Better: Feedback-Efficient Test-Time Adaptation for Vision-and-Language Navigation

cs.CV cs.AI ·2026-10-22 ·Zixu Wang, Huaquan Zhang, Shoukai Xu, Jinwu Hu, Zhiquan Wen +1 more
Abstract

Vision-and-Language Navigation (VLN) requires agents to follow instructions and navigate to target locations. However, current VLN agents often suffer severe performance degradation in unseen environments due to distribution shifts. Test-Time Adaptation (TTA) offers a promising paradigm to bridge this gap, yet existing methods that rely on unsupervised entropy minimization can propagate errors through the trajectory, causing cascading failures in navigation. Recent methods mitigate this by incorporating human feedback, but they apply trajectory-level supervision uniformly across all steps without localizing the actions is truly informative, leading to inefficient feedback utilization and mismatched optimization granularity. To address these limitations, we propose FETA, a Feedback-Efficient Test-Time Adaptation method that improves both feedback efficiency and optimization reliability. To reduce unnecessary human interventions, we introduce a hierarchical sample selection strategy to localize informative samples. Specifically, we first filter trajectories whose normalized policy entropy exhibits large fluctuations. Within each selected trajectory, we further select the critical action via an entropy-gradient saliency criterion and query a single binary feedback on its correctness. A discrepancy-aware entropy optimization objective then updates the policy only at the queried step, with adaptive scaling that assigns larger relative weights to samples with greater prediction-feedback discrepancy. Extensive experiments across multiple benchmarks demonstrate that FETA achieves state-of-the-art performance with fewer human interventions, while remaining robust under noisy feedback and continual adaptation settings.

0.13

When Annotation Patterns Are Learned: Skin-Tone Dependence in Dermatology Concepts

cs.CV ·2026-10-22 ·Y. Li, M. Bigliardi-Qi, S. Bedros, P. Bigliardi, N. Papanikolopoulos
Abstract

Concept bottleneck models (CBMs) let clinicians inspect and correct skin disease diagnoses through clinical concepts, but only if the concept annotations are reliable. We therefore audit the reliability of SkinCon, a widely used dermatology concept dataset, across skin tones. Within the same disease and with near duplicates removed, erythema is annotated far less often on dark than on light skin, with an odds ratio of 0.17. In contrast, hyperpigmentation and hypopigmentation are annotated more often, with odds ratios of 6.2 and 4.5, respectively. We refer to this pattern as tone-dependent concept annotation. An evidential concept model trained on these annotations learns the same pattern and reproduces it on unseen images. Such a model may agree with potential erythema omissions rather than contradict them. Consistent with this, three label error detectors underperform a disease-only baseline at finding erythema that dermatologist captions mention but SkinCon labels as absent. The three detectors are based on confident learning, training dynamics, and the model's own evidential uncertainty. Concept annotations should thus be audited within each subgroup before explanations built on them can be trusted.

0.13

RaSFNet: Ray-aligned and Scan-aware Fusion Network for LiDAR Semantic Segmentation

cs.CV cs.LG ·2026-10-22 ·Chuang Chen
Abstract

LiDAR point cloud semantic segmentation is a fundamental component of 3D perception for autonomous driving and robotics. Projection-based methods map 3D point clouds into 2D representations for feature extraction using 2D networks. However, aggregating multiple points during projection can obscure geometric differences within each frustum, while local feature extraction limits the modeling of spatial relationships across scanning regions. Moreover, proximity in the projected representation does not necessarily imply proximity in 3D space, and interactions between point and projected features may mix features from different ranges, reducing semantic discriminability. To address these challenges, we propose the Ray-aligned and Scan-aware Fusion Network (RaSFNet), which incorporates point cloud geometry and LiDAR scanning characteristics into projection-based semantic segmentation. Specifically, the Ray-Aligned Geometric Distribution Encoding (RAGDE) module characterizes point distributions within each frustum using a ray-aligned coordinate system. To model spatial context, a Scan-Topology-Aware Mamba (STAM) block organizes frustum features according to the LiDAR azimuth sampling pattern and captures long-range dependencies across scanning regions. In addition, a Range-Constrained Point-Frustum Fusion (RCPFF) mechanism uses shared range information to guide feature exchange between point and frustum representations while maintaining spatial correspondence. RaSFNet achieves mIoU scores of 70.2% and 77.7% on the SemanticKITTI and nuScenes validation sets, respectively, indicating competitive segmentation performance compared with the evaluated methods.

0.12

Hear Every Sound Object: Rescuing Overlooked Sources in Auditory Scene Analysis

eess.AS ·2026-10-22 ·Younghoo Kwon, Yongjoon Lee, Jung-Woo Choi
Abstract · 5 pages, 4 figures, Submitted to ICASSP 2027

Auditory scene analysis (ASA) models extract information about individual sound sources from a complex auditory mixture. Recent object-based ASA models process the mixture through a sequence of modules and split it into source-wise features (i.e., object features) only at the last layer. However, substantial disparities in performance across sources limit overall performance. Our analysis using centered kernel alignment (CKA) reveals that CKA scores correlate strongly with per-source performance, with their disparities widening from intermediate to deeper layers. These findings motivate the early split and exchange of object features to address source-wise performance disparities before they widen in deeper layers. To this end, we propose PARASITE, a plug-in refinement architecture that splits object features at earlier layers and progressively refines them through sequential layers. Each layer integrates object features separated at different depths using Object-wise Gated Fusion (OGF) and then refines them using Windowed Cross-Object Attention (WCOA). We introduce the DeepASA++ model, which combines PARASITE with an ASA backbone and delivers state-of-the-art results on both ASA2 and STARSS23. DeepASA++ achieves a SELD score of 0.240 on STARSS23 and surpasses prior multi-stage inference architectures, including chain-of-inference.

0.12

How Depth Growth and Looping Shape In-Context Learning in Language Models

cs.CL cs.AIcs.LG ·2026-10-22 ·Ferdinand Kapl, Kaitlin Maile, Stefan Bauer, Johannes von Oswald
Abstract

Looped and depth-grown language models reuse computations across depth and have attracted interest for their reasoning capabilities. Building on prior suggestive evidence of improved in-context-learning (ICL), we investigate how such reuse favors ICL, including how it shapes the acquisition, retention, and mechanisms of this ability. Our analysis spans pretrained language models up to 3B parameters and controlled tasks. The experiments provide evidence for a shared inductive bias toward learning from context: grown and looped models rely more on information in context and, compared with a full-depth dense reference, acquire ICL earlier in controlled tasks and retain it more strongly when learning fixed associations in the weights is also possible. Tracing attention heads through training reveals that depth growth propagates early induction heads across layers. Ablations show that grown and looped models rely more strongly on induction-ranked heads for both few-shot performance and later-token prediction. Together, these findings provide evidence for a shared inductive bias toward learning from context and suggest that (soft) parameter reuse shapes how this ability develops and which attention mechanisms support it.

0.12

Low-Bit N-Gram Embeddings for Language Models

cs.CL cs.AI ·2026-10-22 ·Stephen Chung, Wenyu Du
Abstract · 24 pages, 10 figures, 10 tables

Many recent open-weight language models use n-gram embedding tables to improve performance. These tables store learned representations of short token sequences, such as pairs or triples of consecutive tokens, which the model retrieves during processing. However, they can require substantial memory: DeepSeek-V4.1-Flash, for example, allocates 196 billion parameters to its Engram n-gram memory. We therefore investigate whether this memory cost can be reduced while preserving the performance improvement that the tables provide. We propose storing n-gram embeddings at low precision and systematically investigate different quantization methods and bit allocations across tables. Across nanochat-, Llama-, and GPT-2-family models trained from scratch, four-bit tables save approximately 75% of table memory while preserving more than 99% of the validation-loss improvement from adding the tables. Using fewer bits leads to a larger drop in performance, although we find that a short fine-tuning phase recovers a significant part of this loss. In addition, reinvesting the saved memory in larger tables can improve performance beyond that of the original full-precision tables while still using less memory. These table memory savings also extend to large existing models: on DeepSeek-V4.1-Flash and Qwen3.8-Flash-Next, low-bit tables save 47-55% of table memory, with perplexity increases of less than 1.2% and largely preserved downstream task performance. These findings establish low-bit precision as a practical way to reduce the memory requirements of n-gram embeddings and use the saved capacity to improve model performance.

0.12

Video-DCI: How Far Are VLM Agents from Human Viewers in Direct Video Corpus Interaction?

cs.CV cs.MM ·2026-10-22 ·Garry Yang, Yuqiao Liu, Hao Lin, Luohao Pan, Xinru Chen +5 more
benchmark
Abstract

Humans answer complex visual questions by inspecting short moments, updating what they know, and redirecting attention toward missing evidence. Existing video benchmarks, however, often provide the relevant video, evaluate only final answers or isolated localization, or permit textual shortcuts, leaving multi-round visual evidence seeking underexplored. We introduce Video-DCI, a diagnostic benchmark and functionally matched human-agent protocol for Direct Video-Corpus Interaction. It comprises DCI-Loc for temporal evidence localization and DCI-QA for end-to-end discovery, retention, and composition of cross-video evidence. In DCI-QA, videos within each pool are topically related, yet every answer requires hidden temporal evidence from multiple videos. Given only the question and video pool, humans and agents iteratively inspect bounded local windows, retain supporting evidence, and answer with cited temporal spans. Human trajectories provide reference curves for evidence acquisition and use across rounds. Experiments reveal a consistent retrieval-to-grounding gap. On DCI-Loc, frontier VLMs often retrieve and expose the reference moment, yet convert far less of this exposure into precise final citations than the human reference. On DCI-QA, the loss begins earlier and compounds across videos. Ablation studies further confirm that iterative search outperforms one-shot retrieval. Multi-round search is therefore necessary but insufficient: current agents accumulate candidates and observations more readily than precise, jointly sufficient evidence.

0.12

H-2DGS: Height-Field Gaussian Splatting for Sharp and Deformable Neurosurgical Surface Reconstruction

eess.IV cs.CVcs.GR ·2026-10-22 ·Zhang Zichuan, Bart Brouwers, Rembrandt Bakker, Maud Landers, Geert-Jan Rutten +1 more
Abstract

Neurosurgical navigation commonly presents magnetic resonance imaging (MRI) as 2D slices, making it difficult to relate the anatomy in those slices to the cortical surface observed through the surgical microscope. We seek a common 3D visualization that combines a photorealistic reconstruction of the exposed cortex during tumor surgery with whole-brain MRI context and diffusion tractography pathways. This requires the reconstructed cortex to form a clear boundary around the brain anatomy rather than a diffuse volume that obscures internal structures. Conventional 3D Gaussian Splatting (3DGS) can reproduce observed views but, under limited surgical view coverage, may distribute opacity through a thick and geometrically ambiguous band. We introduce height-field 2D Gaussian Splatting (H-2DGS), a three-stage pipeline that concentrates a robust 3DGS reconstruction, fuses its depth into a shared truncated signed distance field (TSDF) surface, and represents that surface with curved, surface-oriented Gaussian primitives. Separate height, color, and opacity bases provide local geometry and appearance, while a neighbor-transition objective coordinates overlapping patches. The explicit surface further supports postoperative alignment: MRI-derived depth deforms the reconstruction, and tracked-color supervision with alternating optimization preserves its appearance. Across five real neurosurgical cases, H-2DGS gives low multi-view depth inconsistency, a compact 7.97MB representation, and interactive rendering at 29--42ms per frame. The resulting sharp, deformable surface integrates clearly with tractography and MRI anatomy, including a postoperative resection cavity.

0.11

Spark-Audio 1.0 Technical Report

eess.AS ·2026-10-22 ·Chenyue Zhang, Cong Liu, Dingshu Tian, Dong Sun, Genshun Wan +61 more
Abstract

This paper introduces Spark-Audio 1.0, a speech foundation model trained on domestically produced computing infrastructure. The model accepts text and speech inputs and generates text outputs. It supports multilingual (99 languages) and multi-dialect (202 dialects) speech recognition, speech translation, audio perception and understanding, and both text- and speech-based dialogue, marking a shift from merely ``hearing clearly'' to genuinely ``understanding'' speech. Architecturally, Spark-Audio 1.0 couples a 0.65B-parameter dense audio encoder with a 30B-A3B Mixture-of-Experts (MoE) large language model (LLM). The audio encoder is pretrained with representation-enhanced BEST-RQ and then refined through multi-task supervised training. The audio--LLM is trained through staged pretraining, followed by post-training that combines supervised fine-tuning with on-policy distillation. In total, the model is trained on 13 million hours of audio data together with large-scale text data. We evaluate Spark-Audio 1.0 on mainstream audio and text benchmarks, as well as on in-house test sets drawn from real-world application scenarios. Despite using substantially less audio training data than Qwen3.5-Omni-Flash, Spark-Audio 1.0 achieves comparable performance on most audio and text tasks, outperforms it in multilingual speech recognition, spoken dialogue, and real-world application scenarios, and approaches the performance of larger closed-source models. Nevertheless, there remains room for improvement in instruction following and complex audio reasoning. Finally, the successful training of Spark-Audio 1.0 on domestic hardware, including Huawei Ascend 910B and Hygon BW1000 chips, demonstrates the feasibility of a domestic technical pathway for building speech foundation models.

0.11

Direct cueing of word identity through word duration: evidence from Mandarin homophones

cs.CL ·2026-10-22 ·Xiaomeng Zhu, Megan Niu, Jason Shaw, Kevin Tang
Abstract

While there is increasing evidence that words are stored in the mental lexicon with some phonetic detail, there are open questions about exactly which phonetic details are stored and whether they are functionally relevant for spoken word recognition. Word-level phonetic cues, such as word duration, provide an appropriate test case. Across languages, including Mandarin Chinese, high frequency words are shorter than low frequency words. Using a homophone ordering task, we investigate whether differences in word duration impact word recognition. Mandarin Chinese was examined since it has abundant disyllabic homophones. Our results indicate that participants were able to use word duration to recognize the order of homophone pairs, selected to vary in lexical frequency. Both with high noise levels and when no other cues were available, participants mapped the shorter duration auditory stimulus to the higher frequency word. The results provide evidence that word duration can directly cue the identity of words, suggesting that it may be a phonetic detail incorporated into the lexical representation. We discuss this direct cueing account alongside alternative accounts of how word duration may modulate word recognition.

0.11

Watch and Learn: Teaching Visual In-Context Learning by Removing Textual Shortcuts

cs.CV ·2026-10-22 ·Nimrod Shabtay, Asaf Yehudai, Avihu Dekel, Eli Schwartz, Raja Giryes
Abstract

Vision-language models (VLMs) struggle with in context learning compared to their textual counterparts. Our preliminary study shows that VLMs rely on textual shortcuts and underutilize the visual demonstrations. We therefore hypothesize that VLMs are insufficiently trained to perform visual in-context learning (vICL). To this end, we construct MIME, a task-agnostic vICL training set without textual shortcuts. We ensure the instruction does not state the rule linking the demonstration images to their answers, so the model must infer it from the visual demonstrations. Fine-tuning on MIME improves vICL accuracy by 8.1 to 18.0 points across three VLMs, preserves single-image performance, and improves multi-image and text only in-context learning. In contrast, stating the rule in the training data sharply reduces this gain. We further propose DTAD (Demonstration Text Attention Dropout), which randomly withholds demonstration labels during training so that the VLM must infer the rule from the demonstration images. While achieving comparable vICL accuracy to fine-tuning on MIME, DTAD yields more consistent attention to the demonstration images. On TrueMICL, Qwen3-VL-8B trained on MIME matches frontier models such as Claude Opus 5, and surpasses them with DTAD. Finally, adding MIME to the instruction-tuning mix of LLaVA 1.5 13B more than doubles its vICL accuracy without hurting single-image abilities, whereas a same-size multi-image set improves it only slightly. These results suggest that vICL data should be a standard component of VLM training.

0.10

One Evaluation Is Enough: Inversion-Free Image Editing with Trajectory Correction

cs.CV cs.AI ·2026-10-22 ·Lifan Jiang, Taiyang Chen, Senlin cheng, Bolei Xu, Xiaobo Li +4 more
benchmark
Abstract

Real-image editing commonly relies on inversion and iterative sampling, while efficient one-step editors often stabilize a large update by estimating and averaging editing fields at multiple noise levels. We ask a stricter question: is a single temporal field evaluation sufficient for reliable inversion-free editing? Our answer rests on a previously underused duality in a variance-preserving (VP) one-step model: one source-target query at a shared noisy state simultaneously determines a local angular semantic tangent and a clean endpoint displacement. We integrate the tangent over the analytically known remaining VP angle to obtain a nominal one-step transport, and use the endpoint displacement to correct its long-horizon extrapolation through a constrained projection. This transport-first, correction-second construction yields Endpoint-Calibrated Transport (ECT), a closed-form, training-free editor requiring only one noise level, two text conditions packed into one denoiser batch, and one latent update. ECT requires no image inversion, trajectory integration, second temporal observation, or iterative optimization, and can be directly applied to pretrained epsilon-prediction VP one-step generators. Across the reported benchmark settings, ECT delivers stable gains in editing quality within an extremely short editing time.

0.10

Direct Forcing: Single-Stage Training with Representation-Space Supervision for Few-Step Autoregressive Video Generation

cs.CV cs.AI ·2026-10-22 ·Haojun Xu, Jie Huang, Chuyang Zhao, Xin Lu, Mingchen Zhong +4 more
Abstract

Few-step autoregressive video generation enables low-latency streaming, yet prevailing training pipelines separate causal adaptation from distribution matching. Recent representation-based text-to-video methods retain this separation by starting with an ODE initialization stage, even when the target distribution is defined directly by reference videos. We introduce \method, the first single-stage framework that uses reference videos both as paired examples of how a history continues and as a target population for self-generated rollouts. Specifically, Direct Forcing is trained with two complementary signals: (1) Causal Continuation Alignment learns paired spatiotemporal correspondence under reference histories; and (2) Rollout Distribution Matching employs maximum mean discrepancy to align the representation distributions of self-generated and reference videos. To coordinate these signals, we introduce Dynamic Gradient Balancing, which adapts their relative gradient contributions during training. Experiments with Wan2.1-1.3B show that Direct Forcing learns a four-step autoregressive generator in 250 updates without a separate ODE initialization stage or online diffusion-score supervision, achieving 84.63 VBench Total and 86.01 Quality.

0.10

I$^2$-KAN: Interpretable Intra- and Inter-Modal Multi-Pathway Representation Learning for Multimodal Sentiment Analysis

cs.CV ·2026-10-22 ·Xiaoting Xie, Feng Qiao, Xianwei Zhang, Hailong Chen, Kaiwen Zheng +3 more
Abstract

Multimodal sentiment analysis (MSA) aims to infer sentiment by integrating linguistic, acoustic, and visual information. However, many fusion methods integrate modality-specific representations and cross-modal interactions into joint representations, making it difficult to determine whether predictive utility arises from a modality itself or from its interactions with others. More fundamentally, a modality's standalone predictive utility may not faithfully reflect its interaction-conditioned predictive utility. To address this issue, we propose I$^2$-KAN, an interpretable multi-pathway framework that explicitly maintains intra-modal and pairwise inter-modal information in separate structural pathways. Each pathway is refined by a Task-Decoupled Linear-Gated Denoising Bottleneck (TDL-GDB) to learn compact and task-relevant representations, followed by structured Kolmogorov-Arnold Network fusion that retains explicit pathway associations and supports multi-level interpretation. Extensive experiments on CMU-MOSI, CMU-MOSEI, and CH-SIMS v2 demonstrate strong overall performance. Ablation and interpretability analyses consistently show across the three benchmarks that low standalone predictive utility does not necessarily imply low interaction-conditioned predictive utility: a modality that appears weak in isolation can remain informative through pairwise interactions. These findings highlight the interaction-dependent nature of multimodal utility and suggest that modality utility should be characterized from both standalone and interaction-conditioned perspectives.

0.10

RIFT: KV Cache Compression for Draft Models with Complementary Compressibility Training

cs.CL ·2026-10-22 ·Dor Tsur, Maor Ashkenazi, Maximilian Golub, Peter Hicks, Yaniv Galron +5 more
Abstract

Speculative decoding is a popular decode acceleration method. As such, it is particularly relevant to long-output generation, which requires an increasingly large KV cache. KV cache compression offers a way to constrain costs in long-context drafting. However, compression methods that focus on prefill information can lose relevance as the output length grows. We term this phenomenon decode staleness and show that it degrades draft model accuracy under selection-based KV cache compression. We introduce RIFT, a family of selection-based KV cache compression strategies that balance prefill information and recent generated tokens. We show that refreshing the KV cache selection during token generation significantly improves the drafter's quality at the cost of additional decode latency. To alleviate the refresh costs, we propose an adaptive KV refresh strategy that reduces the number of score calculations while retaining the drafter's quality. We further demonstrate how a compressibility-driven drafter training phase, based on sparse attention masks, can improve the drafter's quality under a compressed KV cache. Combined with compressibility training, RIFT enables Qwen3.5-4B, drafting for Qwen3.5-122B, to retain over 98% of its full-cache acceptance length on 32K-token inputs, 4K-token outputs, and an 8K-token KV cache budget.

0.09

When Do Self-Supervised Representations Help Speech Restoration?

cs.SD eess.AS ·2026-10-22 ·Yongjoon Lee, Younghoo Kwon, Jung-Woo Choi
benchmark
Abstract

Self-supervised learning (SSL) representations are widely used in speech restoration (SR), and the prevailing trend is SSL feature cleansing: fine-tuning the SSL to recover clean features from degraded speech, then conditioning the restoration model on them. However, the representational capability of SSL when encoding degraded speech has been underexplored, and we ask whether cleansing always benefits SR training. Across SSL encoders and perceptual quality proxies, we discover perceptual correlation collapse---under severe degradation, cleansing degraded SSL features toward clean ones misaligns with the SR objective. We hypothesize the restoration backbone can supply the SSL with clues to resolve this collapse, since its intermediate layers already produce partially restored features. We therefore propose Early Restoration Injection (ERI), which routes intermediate features from the restoration backbone into the SSL through a lightweight conditioning network. ERI excels over the unconditioned baseline and, on the VoiceFixer benchmark, achieves competitive performance among multiple baselines.

0.09

Evidence-Guided Selection of Retinal Vessel Segmentation Pipelines: A Cross-Dataset Benchmark of Performance, Structure, Failure Modes, and Reproducibility

cs.CV ·2026-10-22 ·Ruoyu Yan, Yuejia Liu, Shixin Xu, Xiaoyan Li, Xi Luo +2 more
benchmark
Abstract

Retinal vessel segmentation is widely used in retinal image analysis, yet model selection based only on aggregate overlap scores can overlook structural fidelity, cross-dataset robustness, and application-specific error costs. We evaluated five executable binary vessel-segmentation pipelines (FSG-Net, SA-UNetv2, DCP-1024, GAVE, and FR-UNet) on 244 images from six public datasets using pixel-level agreement, cross-dataset stability, thin-vessel retention, structural metrics, error composition, and available computational evidence. Results were aggregated with equal weighting across datasets. No pipeline was optimal in every setting. FSG-Net achieved the highest equal-dataset mean Macro-F1 (0.77), IoU (0.63), clDice (0.83), and Skeleton F1 (0.78), whereas SA-UNetv2 achieved the highest mean sensitivity (0.85) and Thin-Vessel Sensitivity (0.63). Differences across datasets were often larger than those among the leading pipelines, and errors shared by all five methods were concentrated near annotated vessel boundaries and faint peripheral vessels. In an ancillary analysis on 45 HRF images, segmentation masks were converted into 11 ring-wise vascular features for leave-one-out classification. Controlled perturbations showed that local mask errors can be amplified in downstream features, while targeted stabilization reduced sensitivity for five features. The results support task-conditioned, benchmark-relative pipeline selection rather than a single universal ranking and highlight the need to validate the selected pipeline on the intended target domain.

0.09

Squaring-the-Sphere: Distortion-Free and Seamless Panorama Generation via CubeAtlas

cs.CV ·2026-10-22 ·Xuhui Zuo, Chenjie Cao, Tengfei Wang, Junta Wu, Yisu Zhang +1 more
Abstract

Generating 360-degree panoramas requires both local perspective fidelity and global spherical continuity. Existing representations face a trade-off: equirectangular projection (ERP) enables seamless panorama modeling but suffers from severe distortion and a domain gap to pre-trained image generators, whereas cubemaps preserve perspective geometry but often produce inconsistent faces and visible seams. We present Squaring-the-Sphere, a panoramic image generation framework built on a novel representation CubeAtlas, which unfolds all cubemap faces into a single compact canvas. By arranging orientation-aligned, rotated polar regions alongside the equatorial faces, this representation preserves the perspective geometry of cubemaps while organizing the cross-face relationships required for seamless panorama assembly. It further enables a pretrained diffusion transformer to model the complete panorama jointly in a single pass, without altering the backbone architecture or positional encoding. While CubeAtlas has a token count comparable to ERP, its structured layout makes many dense attention interactions unnecessary. We therefore introduce a layout-aware sparse attention mask that preserves the interactions required for panorama continuity while removing geometrically irrelevant token pairs. As a result, CubeAtlas achieves better quality than ERP modeling with lower attention computation. Furthermore, we introduce a training-free super-resolution technique and a fusion scheme that enhance fine details in distortion-free perspective space before achieving the final ERP panorama. Experiments on various benchmarks demonstrate that our method improves geometric fidelity, cross-view consistency, and high-resolution visual quality over previous methods.

0.09

Towards User Simulators That Stay On Task And Still Sound Natural

cs.CL ·2026-10-22 ·Mustafa Omer Gul, Philippe Laban, Tanya Goyal, Jennifer Neville
Abstract · 30 pages, 16 figures, 6 tables. In submission

Users increasingly rely on AI systems for extended, multi-turn tasks, yet evidence indicates that LLM-based system performance degrades over prolonged interaction. To evaluate and improve AI assistants in such settings, we need user simulators that faithfully follow complex user intents while exhibiting realistic conversational behavior. Existing approaches manifest a trade-off: prompted assistants can follow complex user intents but lack behavioral fidelity, while specialized user simulators better approximate conversational behavior but struggle to remain faithful to complex intents. In this paper, we propose a reinforcement learning (RL) recipe that jointly rewards task intent adherence and behavioral naturalness. A crucial contribution of the method is operationalizing naturalness with a lightweight discriminator trained in an alternating, adversarial loop with the user simulator. We release the resulting model, UserLM-2.0, which produces valid simulated conversations (i.e., meeting both task-faithfulness and naturalness criteria) 46.2% of the time, over three times the rate of the strongest prior user simulator (14.1%).

0.08

OneLatent: Latent Reasoning for Efficient Foundation Recommendation Models

cs.IR ·2026-10-22 ·Hao Jiang, Peiru Du, Pengfei Yao, Mengting Li, Siyuan Lou +8 more
benchmark
Abstract

Large language models (LLMs) have demonstrated strong reasoning capabilities, motivating their use as the backbone of foundation recommendation models (FRMs). Existing methods enhance recommendations through explicit Chain-of-Thought (CoT) reasoning under a Think-then-Answer paradigm. However, explicit CoT incurs substantial inference overhead by generating lengthy reasoning traces and relies on manually designed templates that struggle to capture diverse, dynamic user interests. We propose OneLatent, an efficient latent reasoning framework that compresses explicit reasoning traces into several learnable latent tokens, enabling Latent-Reason-then-Answer inference without generating verbose traces. OneLatent first introduces Multi-View Adaptive CoT (MV-ACoT), which creates diverse, high-quality teacher-generated supervision by exploring user interests from multiple perspectives and automatically adapting reasoning complexity to each instance. Building on pretrained FRMs, it then uses a three-stage latent-token alignment paradigm to progressively internalize CoT traces into learnable latent tokens. Finally, a multistage curriculum-based post-training strategy activates latent-token reasoning for downstream recommendation tasks. Experiments on an industrial-scale Kuaishou dataset and the public Kuaishou LLM-Rec benchmark show that OneLatent consistently outperforms explicit CoT-based methods and traditional baselines. Compared with the Think and No-Think variants of FRMs, OneLatent improves SID@64 by 17.44% and 9.33%, respectively, while achieving over 17x higher online inference throughput. We further develop a production serving system for scalable, real-time FRM inference. An online A/B test in Kuaishou's local-services advertising scenario shows that deploying OneLatent with this system yields an estimated 9.6% revenue lift over strong online baselines, including OneRec and OneReason.

0.07

Muting Tokens: Exploring Token Sparsification for Efficient Audio Transformers

cs.SD cs.LGeess.AS ·2026-10-22 ·Shreya Sajid, Fabrizio Julian Piva, Hiram Rayo Torres Rodriguez, Elif Ceren Gok Yildirim, Murat Onur Yildirim +1 more
Abstract · This work was conducted in 2024; the related work reflects the literature at that time

Transformer-based models have shown strong performance across diverse tasks, but their reliance on self-attention mechanisms often leads to high computational costs. Token sparsification, which reduces the number of tokens processed in transformer layers, offers a promising solution to maintain performance while improving efficiency. However, although widely studied in the vision and language domains, token sparsification remains underexplored in audio, with existing work often limited to a single method, model size, or dataset. This work presents a comprehensive evaluation of token sparsification techniques for audio transformers. We assess a range of token pruning and merging strategies across multiple datasets and model sizes, under both post-training and fine-tuning settings. Furthermore, we introduce a novel sparsification approach that combines attention scores with spectral cues derived from the frequency content of token embeddings. Our experiments show that frequency-aware approaches, particularly those based on spectral flatness and low-frequency energy, consistently outperform attention-only methods. We also find that sparsification based on spectrogram patch intensity is especially effective for longer audio inputs. These results highlight the value of incorporating frequency-domain and input-specific information into sparsification strategies to enhance performance in audio tasks.

0.07

LazyFlash: Forward-Dense, Backward-Sparse Attention for Accelerated Long-Context Fine-Tuning

cs.CL ·2026-10-22 ·Keqi Deng, Ruchao Fan, Jinyu Li
Abstract

As large language models (LLMs) process increasingly long sequences, attention becomes a major computational bottleneck. Sparse attention mostly targets inference, yet many methods still train densely or cannot support sparse training, and their added complexity makes sparsity hard to turn into wall-clock speedups, leaving dense attention widely used in practice. This paper introduces LazyFlash, a forward-dense, backward-sparse attention method built on the observation that attention training is dominated by the backward pass. LazyFlash keeps the forward pass exactly dense, so the trained model remains a standard dense model that existing systems such as vLLM can serve without any change. For the backward pass, LazyFlash proposes a gradient-aware block sparsification design that ranks query blocks by their attention-output gradient norms and skips those contributing negligibly to parameter updates, turning the dense backward into a block-sparse one. A gradient-mass budget adapts the amount of skipped computation to each head and training step, while importance sampling preserves the dense gradient in expectation. These designs are implemented inside the tiled FlashAttention backward kernel, where skipped blocks are never loaded or computed, translating gradient sparsity into real wall-clock acceleration. Experiments on long-context text and speech LLMs show that LazyFlash maintains downstream performance while achieving up to 5.7x attention-backward speedup over FlashAttention and 1.7x end-to-end training speedup.

0.07

Free Lunch for LLMs Confidence: The RL Critic Already Knows

cs.CL ·2026-10-22 ·Shujin Wu, Heng Ji
theory
Abstract

Reliable confidence estimation is essential for trustworthy deployment of large language models (LLMs), but dedicated confidence training stage requires additional data and compute. We ask a fundamental question: what prediction problem does confidence training actually solve? Inspired by prior work on external confidence estimation, we compare self- and external estimators of comparable capacity trained on the same (question, response, correctness) triples. We show that an external model trained solely on a policy's behavioral traces can estimate the policy's confidence as accurately as the policy itself, without accessing the policy's internal representations. This points to a deeper connection with value function learning in reinforcement learning (RL): under a correctness-based outcome, we prove that confidence estimation and critic training are equivalent, both predicting the likelihood that a policy's response will be correct from its trajectory. A simple consequence follows: for LLMs trained with value-based RL, the existing critic can be reused as an off-the-shelf confidence estimator with no additional training or data. The policy remains unchanged, so its task performance is preserved, and the critic provides confidence throughout generation, including intermediate reasoning steps. Across math, code, tool calling, and general reasoning spanning a dozen domains, the reused critic is the best confidence estimator on every in-distribution split and the best-calibrated on 14 of 16, without hurting the task performance.

0.06

CAVC: Convergence-Adaptive Audio-Visual Cache for Modality-Asymmetric Diffusion Generation

cs.CV cs.SD ·2026-10-22 ·Lifan Jiang, Yibo Zhao, Senlin cheng, Bolei Xu, Xiaobo Li +4 more
Abstract

Diffusion-based audio-visual generation methods typically perform joint denoising of audio and video latents throughout the entire sampling trajectory, repeatedly executing modality-specific denoising branches and cross-modal attention at every timestep. However, we observe that such uniform joint computation is not always necessary. The key is a modality-asymmetric recoverability: audio-side drift introduced by mid-stage caching can be substantially reduced by only a few terminal joint denoising steps, whereas video-side degradation caused by caching leads to persistent loss of motion and temporal details. This asymmetry suggests that inference acceleration should selectively cache recoverable audio-side computation while preserving full video refinement. Motivated by this observation, we propose \textbf{Convergence-Adaptive Audio-Visual Cache (CAVC)}, a training-free modality-asymmetric caching framework. CAVC adopts a sandwich-structured computation allocation: after detecting that audio denoiser outputs enter a cache-tolerant regime, it caches and reuses the audio denoiser output to bypass audio denoising and cross-modal attention in the middle stage, while the video branch continues independent spatio-temporal refinement; a few joint steps are then restored at the end to reduce accumulated cross-modal drift and improve synchronization. Experiments across multiple public benchmarks demonstrate that CAVC achieves 1.10--1.35x wall-clock acceleration, improves several visual dynamics and temporal statistics, and maintains competitive audio fidelity and audio-visual alignment.

0.06

Loop to Look Better: Learning Adaptive Depth for Question-Guided Perception in Vision-Language Models

cs.CV ·2026-10-22 ·Wencheng Ye, Anning Hu, Hongye Fang, Tianyi Wang, Junchi Yan
Abstract

Visual perception should be iterative and task-directed, with representations refined in response to the question and the need for further processing. While Vision-Language Models (VLMs) increasingly support extended reasoning, their visual tokens are still typically acquired in a single, task-agnostic encoder pass. This creates an evidence-acquisition bottleneck: models may reason longer without refining the visual evidence needed to answer the question. We propose LoopLook, a lightweight encoder-side recurrent perception framework that equips pretrained VLMs with adaptive visual depth. LoopLook recurrently revisits a middle-to-late window of the vision encoder, injecting question representations at each loop to guide visual refinement. A learned stopping mechanism adaptively determines the recurrent depth, while a token-wise Mixture-of-Looks (MoL) combines representations from different loops, allowing each visual token to selectively integrate information acquired across recurrent steps. Across four backbones and eight benchmarks, LoopLook consistently improves performance over the baseline, yielding gains of up to 8.6% on MMStar and 8.1% on HallusionBench. Our analysis shows that LoopLook progressively refines visual attention and adaptively allocates depth according to task difficulty. It also better preserves task-relevant visual information under reduced input resolution, while increasing time to first token (TTFT) by only 5.1-6.5%. By bringing recurrence from reasoning back into perception, LoopLook transforms visual perception from a one-shot glimpse into an adaptive process of recurrent refinement.

0.06

ReflectPO: Reflection-Guided Policy Optimization for Long-Horizon LLM Agents

cs.CL ·2026-10-22 ·Rumei Li, Xunliang Cai
Abstract

With external tools, large language models (LLMs) can tackle complex tasks through sustained environmental interaction. LLM-based agents are therefore increasingly used to tackle complex long-horizon tasks in real-world environments. However, these tasks involve sequential, interdependent decisions and changes in environmental states. Early execution errors affect subsequent interaction histories and environmental states, and their effects accumulate along the decision chain. To address this challenge, we propose Reflection-Guided Policy Optimization (ReflectPO), a reinforcement learning framework for autonomous error recovery in long-horizon tasks. ReflectPO comprises two core components. The first is Reflection-Guided Sampling. We use branch differences and repeated visits in interaction trajectories to construct hints for error recovery. The second is Tree-Structured Credit Assignment. We organize sampled trajectories into trees and assign action-level credit using parent--child reward differences and recovery gains. Experiments over two long-horizon tool-use benchmarks show that ReflectPO outperforms multiple existing baselines in most domains. It also achieves the best overall performance on average. Further analyses show that ReflectPO effectively improves agents' ability to recognize their own deviations, recover from erroneous states, and continue making progress toward task completion.

0.05

Security & Privacy 29

MIG-COW: Transferable Adversarial Attacks on Deepfake Detectors via Gradient Decomposition

cs.CR cs.AIcs.LG ·2026-10-22 ·Wonjune Seo, Joonhyuk Baek, Yeseong Jung, Saerom Park
top venuehighly cited
Abstract · 7 pages, 3 figures, ACM MM '25

Despite recent advances, deepfake detectors remain vulnerable to adversarial examples, particularly in diverse, real-world settings. We propose MIG-COW, a novel adversarial attack framework that generates highly generalizable and visually imperceptible adversarial examples. By combining momentum-integrated gradients with a consensus-orthogonal decomposition, MIG-COW captures both shared and model-specific vulnerabilities across heterogeneous CNN and ViT detectors. On the AADD-2025 Challenge benchmarks, MIG-COW achieves a 99.96% white-box attack success rate (ASR) with high perceptual similarity (SSIM), significantly outperforming existing baselines. However, its limited 7.16% ASR against official black-box targets - despite achieving the best overall score - highlights the ongoing challenge of transferability. We also demonstrate that incorporating low-performing but diverse models in the ensemble can degrade attack effectiveness, underscoring the need for careful surrogate model selection in real-world adversarial settings.

0.64

Agentic AI Defense: Decode-Time Provenance Binding for MCP

cs.CR cs.AIcs.MA ·2026-10-22 ·Eric Yocam, Varghese Vaidyan, Paris Kalathas, Patrick Tague, Bruce DeBruhl +1 more
code/data
Abstract · 12 pages, 5 figures, 13 tables, 2 algorithms. Code and data available at https://github.com/ericyoc/stego_provenance_defense

Agentic AI systems using the Model Context Protocol (MCP) are vulnerable to indirect prompt injection through untrusted tool outputs. We demonstrate cross-modal trust laundering, in which instructions concealed in an email-signature image using least-significant-bit (LSB), discrete cosine transform (DCT), or metadata steganography bypass text scanners and enter a multi-agent workflow when decoded by a tool. To counter this threat, we propose decode-time provenance binding, which combines HMAC-authenticated trust labels with an MCP-maintained provenance ledger and context-taint tracking. The mechanism labels decoded content as untrusted and blocks critical tool calls influenced by that content, even when a compromised agent substitutes a valid trusted label. We evaluate the approach using a real MCP testbed with three steganographic carriers and 24 adaptive payloads. The undefended pipeline exhibits 100% attack success, while keyword and regex defenses each permit 62.5% of attacks. Our mechanism blocks all attacks (0% success), including adversarial label substitution across mediated agent boundaries. A four-component ablation validates the roles of decode marking, cryptographic authentication, lineage tracking, and sink enforcement. An audited trusted-promotion mechanism increases legitimate-task completion from 67% to 100% without compromising protection. The approach provides carrier-independent containment within fully mediated MCP execution environments. Code and data are publicly available.

0.58

AfterDeleteBench: Auditing Post-Deletion Information Persistence in LLM Agents

cs.CR cs.CLcs.LG ·2026-10-22 ·Ehsan Saleh, Roozbeh Nahavandi, Sama Salarian, Raef Bassily, Andrew Perrault
Abstract · 25 pages, 7 figures, 12 tables

Many LLM agents maintain external persistent memory and derive additional state from past interactions, such as summaries. Deleting an original memory record therefore need not remove information carried by its descendants. We introduce AfterDeleteBench, a controlled behavioral audit for this failure mode in fixed-weight LLM agents. For each synthetic secret, we compare an exposed-then-deleted agent with an otherwise matched agent that was never exposed to the secret, allowing post-deletion recovery to be attributed to the original exposure. In our primary setting, a single query recovers the deleted secret from Qwen2.5-7B-Instruct in roughly two thirds of cases, while matched never-exposed controls have zero recovery. Crucially, recovery does not require a surviving verbatim copy: when the complete secret string is absent and only separately stored components remain under a public composition rule, exact reconstruction succeeds in 62-99% of cases across Qwen, Mistral, and Phi-3. We also show that the response interface is part of the measurement system: target-bearing replies can be scored as failures when the requested format and parser are misaligned. Finally, we separate reconstruction from write-back and fresh-session recall. Provenance-aware closure eliminates observed recovery in the tested positive-baseline settings, whereas an exact-literal write filter blocks reinsertion but cannot erase information already present. Together, these results show that source deletion should be evaluated behaviorally at the level of the agent and its memory system, not only as a storage operation.

0.47

Incomparable Minima in Agent Tool Authorization

cs.CR ·2026-10-22 ·Xin Xu
published
Abstract · Accepted as an oral at TAS 2026, AAAI Fall Symposium Series. 9 pages

An agent that needs a tool needs authority to call it, and every proposal for governing that authority has to answer one question: of two grants that both let the agent finish a task, which one is smaller? Over-privilege benchmarks score agents against this answer, and permission solvers compute it. We check the question against the authorisation tables that Microsoft and Google publish, covering 22,305 Microsoft Graph endpoint bindings and 23,852 Google API methods, on tasks generated from each table's own resource structure. For 29.8-33.3% of Graph tasks and 3.2% of Google resource tasks the question has no answer, because the minimal sufficient grants form an antichain, a set of grants none of which authorises a subset of what another does. In its two main authorisation modes, Microsoft's shipped privilegeLevel rates 67.4-74.4% of these incomparable pairs equal. Google ships no such scale, so on Google we apply a reconstruction of the FORTIS privilege ladder, which rates 90.7-98.7% equal. Applying the same reconstruction to Graph attributes much of that difference to the scale rather than the corpus: its form based on the HTTP verb alone rates 92.0% of Google pairs and 89.0-92.0% of Graph pairs equal. Four risk measures (operations, writes, services reached, administrative surface) order 28.7% of incomparable Google pairs and 15.7-23.4% of Graph pairs in opposite directions, so for those pairs the ranking has no measure-independent answer. Offered pools of minimal and larger grants, 6 frontier models pick a minimal grant on 99.0% of 1,708 decisions, a grant smallest under all four measures on 66.7%, and on 22.1% a grant that another option matches or beats on every measure. We propose that benchmarks report which minimal grant an agent chose and that deployments show the alternatives to whoever approves the grant.

0.46

BelT Block Cipher

cs.CR ·2026-10-22 ·Sergey Agievich, Victor Galinskij, Nikolai Mikulich, Yuriy Kharin
publishedunusual
Abstract

BelT is a block cipher with a 128-bit block size and a 256-bit key, developed in Belarus and adopted as part of the Belarusian national standard STB 34.101.31 for encryption and integrity control. Its specification was issued as a preliminary standard in 2007, becoming a final standard in 2011. BelT employs an eight-round structure combining elements of a Feistel network and the Lai-Massey scheme. This paper presents a complete description of the cipher and explains the principles underlying its design. Software performance and possible optimizations of BelT are discussed in comparison with those of candidates in the Advanced Encryption Standard (AES) competition.

0.43

ReDence: A Hybrid Agentic and Decision-Model Framework for Rapid, Evidence-Grounded Detection of Modern Social Engineering Scams

cs.CR cs.LG ·2026-10-22 ·Jaishnoor Kaur, Josef Horáček, Ian Waskom, Sayak Saha Roy
code/data
Abstract

Detecting online social engineering (SE) scams often requires examining website workflows, verifying claims against external sources, and connecting evidence across multiple pages. Existing detectors provide limited support for these investigations, leaving analysts to manually gather and validate evidence, which inevitably delays detection and subsequent takedown of these attacks. To address this gap, we present ReDence, an open-source hybrid framework for adaptive, evidence-grounded detection across diverse SE scam categories. ReDence combines two locally hosted models with complementary responsibilities: a fine-tuned multimodal agentic Investigator (Qwen3.5-4B) that performs investigative planning, evidence interpretation, and reporting, and a specialized Decider (Tev1) that handles binary and multiple-choice tasks, including browser-action selection, evidence assessment, and decisions about whether to continue an investigation. By delegating discrete decisions to a specialized model, ReDence reduces reliance on lengthy language-model responses, while preserving the reasoning capabilities required for complex investigations. Our evaluation demonstrates that ReDence achieves 96\% accuracy in detecting scams across twelve SE scam categories, substantially outperforming existing detectors while being at least four times faster than comparable agentic models without compromising investigative depth. ReDence also produces structured, evidence-grounded reports that document investigative actions, supporting and contradictory evidence, and collected artifacts, which allow analysts to independently audit detection decisions and utilize the collected evidence to support subsequent mitigation and takedown efforts.

0.42

Your Token Is Gone: A Token-Consumption Countermeasure for Disrupting LLM-Based Scam Systems

cs.CR ·2026-10-22 ·Fudu Xing, Kailong Wang, Tianlong Yu
Abstract · 10 pages, 5 figures, 3 tables

LLM-based scam systems increasingly combine adaptive dialogue, persistent context, external tools, and multi-agent coordination to automate personalized fraud. These capabilities also create a resource dependency: every reasoning step, model call, and coordination round incurs inference cost. We present TokenTrap, a resource-oriented defense that redirects suspected scam systems toward benign, low-value computation while withholding credentials, payments, and other exploitable information. TokenTrap combines three components: an architecture-specific token-task library, a router that selects among single-agent, hierarchical, and flat multi-agent workflows, and a bounded anti-guardrail strategy that varies task presentation while enforcing token, round, time, and safety limits. Across 50-round traces, TokenTrap amplifies token consumption by $2{,}487.0\times$, $1{,}616.0\times$, and $567.8\times$ relative to a no-token-task baseline for the three workflows. Its router achieves 85.67% accuracy on 300 architecture-labeled prompts. These results demonstrate that short, non-sensitive requests can impose substantial context-processing and coordination costs, motivating resource-aware defense as a complement to scam detection.

0.41

SILMARILS-MV: Information-Theoretically Secure Multi-Verifier Signatures with Dealer-Free Key Issuance and Linear Aggregation for Permissioned Blockchains

cs.CR cs.DCcs.ITquant-ph ·2026-10-22 ·Khadijeh Bagheri, Hassan Khodaiemehr, Chen Feng
Abstract · 24 pages, 8 figures

Permissioned blockchains authenticate transactions, votes and quorum certificates with computationally secure signatures, and post-quantum replacements inflate consensus traffic by orders of magnitude. Information-theoretic alternatives have required a trusted authority, quantum channels, per-signature interaction, or single-use keys with bounded transfer. We present SILMARILS-MV, a multi-verifier unconditionally secure signature in which a signature is the vector $α= A\,e(m)$ and each verifier holds a secret functional $c_i = Γ^T λ(x_i)$ derived from virtual anchors $Γ$ that nobody learns. Against unbounded adversaries corrupting up to $ω$ verifiers and any signers, and observing every decision, an honest verifier accepts a forged or invalid signature with probability at most $Q/q + 1/(q-1)$, a bound that is tight in its leading term. The scheme is strongly unforgeable, non-repudiable, transferable across unboundedly many hops, and linearly aggregatable. Keys are issued by the validators through perfectly secure MPC for $t < n/3$, without broadcast. To scale, we let the committee itself sign: in threshold notarization, finality certificates have constant size $32(ω+1)$ bytes and per-validator issuance cost no longer depends on $n$. Arbiter-verified keys give clients 32-byte signatures, robust redistribution supports membership changes and share refresh, and verifiers can be enrolled retroactively. A certificate verifies in 2.4 microseconds. In an analytic normal-case model at 1000 validators, threshold notarization needs 71 KiB of authentication traffic per validator per block, 22x less than ML-DSA-44 for admitted verifiers. In a Rust HotStuff prototype with 100 validators, SILMARILS-MV commits 2.8x more transactions per second than ML-DSA-44 with 67x less cryptographic CPU, and Monte-Carlo attacks match the bounds.

0.35

SecLog-Bench: A Scalable Benchmark for Cybersecurity Log Reasoning with Scaffolded Evidence

cs.CR cs.AI ·2026-10-22 ·Yifan Wang, Dongliang Xu, Xiaoyong Li, Jing Li, Peter S. Pan +1 more
benchmarkcode/data
Abstract · 17 pages, 9 figures. Submitted to IEEE Transactions on Dependable and Secure Computing. Code and data: https://github.com/SecLog-Open/SecLog

Large language models (LLMs) have shown strong potential for cybersecurity, leading to growing interest in LLM-based automated cybersecurity log analysis. However, current datasets (1) are limited in scale; (2) rely mainly on single-source evidence; and (3) have only binary or multi-class attack labels, without dense annotations of the reasoning process. To address these limitations, SecLog-Bench has three key features. First, we build an extensible simulation platform that covers 29 attack scenarios, supports repeated execution and the integration of new scenarios, and enables scalable generation of cybersecurity logs with attack-stage labels. Second, we use multi-source logs, including audit logs, application logs, and network traffic, to jointly analyze cyber attacks. Third, we annotate each sample with the essential context for cybersecurity log reasoning and a structured step-by-step evidence chain. We evaluate eight mainstream LLMs on this benchmark for post-detection attack-progress reasoning. Results show that, when provided with scaffolded evidence, current LLMs exhibit modest capability in attack-progress reasoning, with the best model achieving only 44.44% accuracy and 45.07% Macro-F1. Their reasoning remains unreliable; i.e., the highest Stage Reasoning F1 (SRF1) score is only 54.87%. Moreover, even correct predictions may be accompanied by flawed reasoning, and reasoning quality is often misaligned with prediction accuracy.

0.35

FVQKD: A Formal Verification of Quantum Key Distribution Security in Rocq

cs.CR quant-phcs.LO ·2026-10-22 ·Ruixuan Deng, Zhengfeng Ji, Jiayu Zhang, Li Zhou
code/data
Abstract · 87 pages: 12 pages of main text, 4 pages of references, and 71 pages of appendices. Artifact: https://github.com/fvqkd/FVQKD

Quantum key distribution (QKD), whose security is derived from quantum mechanics rather than computational assumptions, is moving into standardization and commercial deployment. Yet its security guarantees rest on technically delicate proofs checked by hand. Formal verification closes this gap, establishing a two-factor assurance: unconditional security from physical principles, and machine-checked proofs against reasoning errors. We present FVQKD, a Rocq library for machine-checking QKD security proofs. Its first result is a complete, end-to-end machine-checked proof of composable, finite-key security for BB84. The mechanization reconstructs the complete proof infrastructure from first principles, including the smooth entropy framework, the entropic uncertainty relation, and the Quantum Leftover Hash Lemma. It uncovers necessary side conditions that the published argument leaves unstated, one of which reveals a missing proof branch, and it grounds the information-theoretic bound in a formal semantics of protocol execution, which the pen-and-paper analysis lacks. The proof is entirely admit-free, relying only on the Rocq kernel and standard foundational axioms. To our knowledge, this is the first end-to-end formal verification of a QKD security proof. Built on MathComp, MathComp-Analysis, and CoqQ, the development provides quantum-information infrastructure designed for reuse across broader QKD variants and quantum cryptography applications.

0.33

Structural cryptanalysis of classic McEliece: Assessment of claims, costs and evidence

cs.CR ·2026-10-22 ·Ramses Fernandez
Abstract

For the last two years, classical structural attacks on McEliece with binary Goppa codes progressed further, at the parameters of Classic McEliece, than in the previous four decades: a subexponential and then a quasi-polynomial distinguisher, key recoveries for public challenges with brute-force labels up to $254$, and conditional key-recovery estimates of about $2^{89}$ to $2^{128}$ per run of a large sparse linear-algebra computation, depending on the parameter set and on whether and how memory access is charged. On September 2026 the German BSI advised against new deployments of the scheme. This paper gives a version-aware critical assessment of these results. It separates proved distinguishers, conditional cost estimates and actual recoveries; analyses in detail Saarinen's HOVER algorithm, his conditional cost analysis and his recovery of the TII-254 challenge, together with Weis's cost study of Holdout/Hermite key recovery; and compares the other key-recovery routes and the objections raised against them. Our own analysis finds that the cost model is now the main quantitative uncertainty, with about thirty bits between the cheapest and the most expensive model for the same computation; that for the four larger parameter sets every model considered leaves the conditional estimates far below the targeted levels, whereas for mceliece348864 the shortfall is of the order of the spread between models; that the structural estimates are lower than Grover-accelerated information-set decoding for the four larger sets in essentially every model and comparable for mceliece348864; and that none of this is a demonstrated attack, since the underlying heuristics have been tested only on instances tens of thousands to hundreds of thousands of times smaller than the target. We close with open questions.

0.32

Weaponizing the Exit: Input Privacy Leakage via Exit Routing Manipulation in Multi-Exit Networks

cs.CR ·2026-10-22 ·Yulong Tian, Lei Wang, Jincan Yang, Hao Wu, Hao Han +2 more
code/data
Abstract

Multi-exit networks improve inference efficiency by allowing inputs to terminate early at intermediate exits. We identify a privacy vulnerability in cloud deployments of these networks: a malicious Cloud Service Provider (CSP) can infer sensitive properties of individual user inputs solely from inference latency, without accessing inputs, intermediate states, or model outputs. The attack exploits the CSP's control over the model supply chain. Prior to deployment, the CSP supplies a specially crafted multi-exit network manipulated so that inputs carrying a target sensitive property are routed to a designated exit with high probability, while other inputs exit elsewhere. At serving time, the CSP infers the exit index from per-query latency, thereby determining whether each input possesses the target property without accessing inputs, outputs, or internal model states. We propose a training framework combining selective label smoothing and entropy-covariance optimization to control confidence distributions across exits and reduce premature termination. We conduct experiments on four datasets across ten target properties. The results demonstrate strong attack effectiveness and stealthiness. The attack achieves an average F1 score of 0.73, compared to 0.25 for the strongest baseline, while main task accuracy drops by less than 1.3\% and inference speedup degradation remains below 0.1 in most settings. Our code is available at https://github.com/kazefjj/Weaponizing_the_Exit.

0.30

Practical Auditing of Differentially Private Graph Release Mechanisms: Methodology, Toolkit, and Case Study

cs.CR ·2026-10-22 ·Nicholas D'Silva, Surya Nepal, Salil S. Kanhere
publishedcode/data
Abstract · Accepted to AISC '27

Differential Privacy (DP) has become a prominent approach for protecting graph data releases, yet empirical auditing methodologies have largely been developed for tabular datasets and machine learning pipelines rather than graph-structured outputs. We present a methodology for empirically auditing DP graph release mechanisms, identifying two graph-specific audit design problems: canary construction and witness statistic selection. We combine these components with an established statistically certified testing procedure and implement the resulting methodology as an open-source Python toolkit. To demonstrate the approach, we audit a published state-of-the-art Node-DP graph release mechanism. Across a series of synthetic and real-world canary pairs, we obtain certified lower bounds on the privacy parameter {\varepsilon} that exceed the claimed value, yielding counterexamples sufficient to falsify the claimed guarantee for the audited implementation. The violations persist under progressively stronger restrictions on the input domain, including non-empty and connected graphs. We also show that witness selection can materially affect whether a violation is detected, highlighting the importance of evaluating a diverse battery of statistics when auditing DP graph release mechanisms.

0.30

TRACE: Data-Free Text Reconstruction Attacks against Approximate Unlearning in LLMs

cs.CR cs.LG ·2026-10-22 ·Mengying Zhang, Derui Wang, Ibrahim Khalil, Bowen Liu, Xiaoyu Xia +1 more
top venue
Abstract · Accepted to The Fortieth Annual Conference on Neural Information Processing Systems (NeurIPS 2026)

As large language models (LLMs) face growing demands for data removal driven by regulatory compliance, copyright concerns, and privacy protection, approximate unlearning has emerged as a practical alternative to expensive full retraining. In practice, pre- and post-unlearning model snapshots are often maintained for auditing and rollback, introducing a previously overlooked side channel. This paper demonstrates that approximate unlearning in LLM can leave residual traces within model snapshots, which may be recoverable and thus enable the privacy breaches that approximate unlearning is intended to prevent. Here, we propose TRACE, a framework that reconstructs unlearned training data given access to pre- and post-unlearning model snapshots, without any data-specific prior knowledge. To achieve this, TRACE identifies a compact set of candidate tokens from embedding-level parameter differences to constrain the search space, and assembles sequences via a contrastive scoring mechanism based on the perplexity gap between the two models. A two-stage exploration-then-completion strategy enables the recovery of multiple distinct samples. Extensive experiments on various datasets, including TOFU, MUSE, and WMDP, across multiple LLMs and six representative unlearning methods, demonstrate that TRACE achieves high-fidelity reconstruction. This paper exposes significant privacy risks in current LLM approximate unlearning deployments and highlights the need for defenses against parameter-difference leakage and appropriate management of model snapshot access.

0.30

SEMPRE: Semantic-Aware Protocol Reverse Engineering via Consistency-Driven Joint Optimization

cs.CR ·2026-10-22 ·Yihua Xu, Weijie Huang, Xiaofeng Wang, Hang Zhang, Binghui Wang +1 more
Abstract

Protocol reverse engineering (PRE) is essential for understanding undocumented or proprietary communication formats, enabling security-critical tasks such as vulnerability discovery, malware C2 analysis, and IDS signature generation. However, existing PRE techniques rely on brittle hand-written rules, under-model contextual dependencies among neighboring fields, and often fail on messages with nested sub-protocols. We present SEMPRE, a fully automated PRE framework that replaces ad-hoc heuristics with recursive LLM reasoning grounded in fine-grained dynamic program analysis. SEMPRE uses taint-flow and loop traces to guide an LLM-driven tree-of-thought search that slices messages into fields while detecting nested sub-payloads, producing a hierarchical layout. It then infers each field's type and role and iteratively refines boundaries by freezing high-confidence segments and re-running detection under semantic constraints, yielding an EM-style procedure that monotonically improves structure and semantics until convergence. Evaluated on a broad corpus of mainstream protocols against six state-of-the-art (vs. <0.4 for all baselines) and 0.72 for semantic inference (vs. 0.13 for BinPRE). Ablations confirm the contribution of each component, and a case study on a prolific real-world botnet demonstrates practical utility.

0.29

Bridging the Closed-World Gap: Interpretable Anomaly Detection via a Hybrid TSK-FCM Ensemble on the BETH Dataset

cs.CR cs.AIcs.LG ·2026-10-22 ·Lohith Srikanth Pentapalli, Josette Riep, Justin Ouwerkerk, Kelly Cohen
publishedbenchmark
Abstract · Presented at North American Fuzzy Information Processing Society (NAFIPS) 2026, El Paso, Texas, USA; accepted for publication in Springer pr…

A fundamental paradox plagues the deployment of Machine Learning in cybersecurity: supervised models generalize poorly to novel, Out-of-Distribution (OoD) attacks, whereas unsupervised models often suffer from poor separability, leading to unmanageable false-alarm rates. This challenge is exemplified by the BETH dataset, a kernel-level benchmark in which the training data consists solely of benign and suspicious insider behaviors, whereas the test set consists almost entirely of novel botnet activation attacks. To bridge this detection gap, we propose a Hybrid Fuzzy Ensemble that synergizes the precision of supervised learning with the adaptability of unsupervised clustering. Our architecture integrates a gradient-optimized Takagi-Sugeno-Kang (TSK) fuzzy classifier to act as an expert on known insider threats, with a parallel Fuzzy C-Means (FCM) module to detect geometric anomalies. A critical finding of our research is that traditional fuzzy membership scores degrade in high-dimensional feature spaces due to the curse of dimensionality. We overcome this by introducing a robust Min Distance metric that exploits the raw geometric isolation of botnet activation attacks. This approach yields a state-of-the-art AUROC of 0.9951 on the BETH test set, outperforming the original Isolation Forest baseline by 17.1%. By combining near-perfect OoD detection with the intrinsic interpretability of fuzzy logic, our framework offers a viable path toward robust, transparent, and autonomous network defense.

0.29

Bridging Number-Theoretic Primality Testing And Algorithmic Optimization: An Adaptive Cryptographic Framework

cs.CR cs.LG ·2026-10-22 ·Tran Minh Thanh, Vo Hoai Viet
Abstract

The security of modern public-key cryptographic systems, such as RSA, Diffie-Hellman, and ECC, depends entirely on the computational intractability of the integer factorization and discrete logarithm problems. For these asymmetric algorithms to operate, a core requirement is that the system must be capable of generating, testing, and simultaneously proving the primality of extraordinarily large numbers. To meet those requirements, this study proposes a comprehensive framework, bridging clas- sical number theory with modern machine learning models and optimization algorithms. Specifically, the paper delineates an adaptive, arbitrary-precision architecture dedicated exclusively to primality testing. Concurrently, the research deeply analyzes the theoretical foundations of testing methods (both deterministic and probabilistic), hardware-level optimization techniques, and the application of quantum-secured entropy sources, while providing rigorous mathematical proofs to guarantee security even when the prime distribution is biased.

0.27

Social-HUMINT: An LLM-Augmented Unified Framework for Measuring Social Media Exposure

cs.CR cs.AIcs.SI ·2026-10-22 ·Pellegrino Casoria, Simon Pietro Romano
hot direction
Abstract

In the post-API landscape, social-media profiling has become inexpensive and automated: Large Language Models (LLMs) can extract structured information from unstructured content and deduce attributes users never disclosed. A person's true exposure is not what they post but what can be inferred from it. As these practices become routine, exposure to social engineering attacks becomes an operational certainty, demanding a shared framework for assessment and mitigation. We make this exposure measurable by quantifying how accurately a target can be profiled, and how far that profile enables social engineering attacks. Because no existing model fully spans this continuum, we unify Social Media Intelligence (SOCMINT) and Cyber Human Intelligence (Cyber-HUMINT), two individual-focused disciplines that evolved separately, into a single doctrine-grounded taxonomy of 38 capabilities across the Cyber Threat Intelligence lifecycle, where user profiling bridges passive collection and active engagement. Domain experts validated the taxonomy (Fleiss' $κ$ = 0.92). We operationalize its core path (extraction, inference, and susceptibility assessment) as an LLM-augmented multi-agent pipeline, evaluated on real participants' public data. Susceptibility is assessed indirectly: the pipeline predicts how each target would rank personalized engagement options, matching their actual choices significantly above chance. This ranking capacity captures the personalization on which tailored attacks depend. The underlying profiles are accurate: extraction reaches F1 = 97.3%, and inferred attributes score 4.73/5, both far above rule-based and NER-based baselines at negligible cost. The framework augments security analysts' tradecraft rather than replacing it and enables defenders to gauge their own exposure before an adversary does.

0.27

AgentPhishing: A Benchmark of LLM Email Agents Under Adaptive Attack

cs.CR cs.AI ·2026-10-22 ·Justin Tan, Miao Li, Ding Zhao
benchmark
Abstract · 28 pages, 6 figures

Computer-use agents increasingly handle private information and act on users' behalf across desktop and web applications, including email. Phishing therefore poses privacy and security risks that depend not only on whether agents recognize malicious messages, but also on how they respond. Existing phishing datasets were developed primarily for conventional detection, leaving agent susceptibility to persuasive malicious requests insufficiently characterized. To expose vulnerabilities that pre-existing datasets may miss, we introduce AgentPhishing, a benchmark for evaluating whether persuasive phishing attacks can evade detection and induce LLM email agents to comply with malicious requests. We develop PhishOpt, which optimizes reusable email-generation skills against individual agents under fixed benchmark constraints and a realism constraint. We analyze the generated emails using Cialdini's persuasion principles to identify strategies associated with agent failures and use these findings to construct a universal generation skill. Our evaluation shows that agents can handle pre-existing phishing emails safely yet remain susceptible to attacks adapted to agent behavior, while in an AgentPhishing evaluation of nine agents, four exhibit phishing-detection rates below $10\%$ and malicious-request compliance rates above $45\%$. These results highlight the limits of inferring agent safety from conventional phishing evaluations and the need to assess how agents respond to persuasive malicious requests.

0.26

Ulysses: A Security Harness For Synthesizing Verifiable Web Exploits

cs.CR cs.SE ·2026-10-22 ·Aditya Oke, André Vicente Duarte, Rui Melo, Shubham Gandhi, Nachiket Kotalwar +4 more
benchmark
Abstract · 23 pages, 6 Figures

How can an AI agent automatically discover security vulnerabilities in any software repository? Existing approaches use LLMs to generate software vulnerability exploits. However, it is challenging for an LLM to reliably verify its own generated exploits. In this paper, we propose Ulysses, a multi-agent security harness to synthesize verifiable vulnerability exploits for web applications. Ulysses converts candidate vulnerability findings into verified Proofs of Exploitability by iteratively exploring exploit hypotheses and establishing success in independently observed execution effects. We evaluated Ulysses on WebExploitGym, a curated benchmark of exploitable instances from 25 real-world web applications that span 5 languages and 7 frameworks. Ulysses produces verified proofs for 10 of 25 real-world advisories. It also discovered 4 new proofs of exploitability that have never been previously reported.

0.25

Model-intrinsic, Weight-level, and Training-free Spectral Fingerprinting for Large Language Models

cs.CR ·2026-10-22 ·Yanan He, Yanli Chen, Ling Xiong, Gejian Zhao, Hanzhou Wu
Abstract · To appear in IEEE International Workshop on Information Forensics and Security (2026)

The increasing release and reuse of Large Language Models (LLMs) as open-weight checkpoints make model fingerprinting important for ownership protection and offline auditing. Existing fingerprinting methods typically construct evidence from query responses, generated outputs, or input-induced activations. Although effective in representation-based settings, they can be affected by prompt selection, decoding configuration, inference environment, and forward-pass implementation, which may limit their applicability. Here we introduce a model-intrinsic weight-level framework that derives compact fingerprints from spectral structures encoded in LLM parameters. Specifically, the framework constructs GATE spectral scalar features from the singular-value distributions of GATE projection weights in gated feed-forward networks (FFNs), and aggregates the resulting layer-wise features through depth-wise segmented pooling to form a model-level fingerprint. Such representation preserves coarse layer-depth organization while reducing dependence on strict layer-to-layer alignment, enabling comparison across LLMs with different depths. The similarity between fingerprints is computed by linear centered kernel alignment. The entire pipeline is training-free and does not require input samples, generated outputs, forward activations, or an auxiliary verifier. Experiments on representative LLM families and post-training variants have demonstrated a clear separation between source-related models and unrelated ones, and stable similarity under instruction tuning, quantization, pruning, and distillation-related modifications, which verify that spectral features are robust evidence for LLM fingerprinting.

0.25

Omni-UBench: Benchmarking Omni-Modal LLM-based Guardrails Across Safety, Utility, and Reliability

cs.CR cs.CLcs.HC ·2026-10-22 ·Junhao Wang, Yidou Chen, Li Lu, Hongwei Yao, Xiaodan Li +4 more
benchmark
Abstract

Advances in Omni-modal Large Language Models (OLLMs) have enabled increasingly seamless human-AI interaction, but their unified processing of heterogeneous modalities also expands the attack surface, allowing adversaries to exploit richer input compositions to circumvent safety alignment. To counter such risks, omni-modal guardrails have been proposed to detect and filter unsafe queries across arbitrary modalities. However, their capabilities and limitations remain insufficiently understood due to the absence of a comprehensive evaluation benchmark. In this paper, we present Omni-UBench, the first comprehensive framework for omni-modal safety assessment, evaluating guardrail models along three critical dimensions, i.e., safety, utility, and reliability. Omni-UBench curates over 183k samples drawn from 35+ heterogeneous sources spanning text, image, audio, video, and their cross-modal combinations, annotated with safety labels under a unified risk taxonomy. Benchmarking under Omni-UBench on 15 representative OLLMs and omni-modal guardrails yields several key findings, including: (1) modern OLLMs exhibit strong content moderation potential even without guardrail-specific tuning; (2) existing guardrails still struggle to identify cross-modal queries that conceal harmful intent or involve sophisticated jailbreaks; (3) they further exhibit substantial safety hallucination and over-sensitivity to textual patterns, revealing a deficient understanding of cross-modal safety semantics. Our findings provide valuable insights for developing more trustworthy omni-modal guardrails in future research.

0.24

AIRO: A Streaming-Based Machine Learning Framework for Threat Detection and Security in Industrial Internet of Things

cs.CR cs.LGcs.NI ·2026-10-22 ·Raima Deb, Pranav M Pawar, Mithun Mukherjee, Raja Muthalagu, Mian Guo
Abstract · 7 pages, 6 figures

Modern Industrial Internet of Things (IIoT) im- plementations now connect edge gateways, cloud dashboards and business applications. This improves visibility and remote monitoring. However it simultaneously exposes industrial traffic to various cyber threats including scanning, web attacks and Structured Query Language (SQL) injection (SQLi), phishing- driven credential theft, and command-and-control (C2) activity. Although rule-based tools remain useful for detecting known signatures, their efficacy decreases with minimally altered attacks or when alerts require manual investigation. This paper presents AIRO (Autonomous Intelligence-Driven Cybersecurity Risk Re- duction System), an efficient framework for IIoT monitoring and response. AIRO combines Apache Kafka streaming, attack- specific feature extraction, machine learning (ML) inference, confidence-based decision logic and risk reduction measures such as Internet Protocol (IP) blocking, account suspension and traffic limiting. We developed four specialized datasets for C2 beaconing, reconnaissance, SQLi, and phishing to evaluate AIRO. These datasets include noisy and overlapping characteristic distribu- tions to minimize label leakage. The best-performing models, primarily Gradient Boosting and LightGBM, achieve F1-scores up to 0.9229 while ensuring low-latency from event input to mitigation

0.23

Auditing the Hand-Written Inputs of Learned Access Control

cs.CR cs.AI ·2026-10-22 ·Arsalan Jalil Sarghaleh, Mehdi Hazrati
Abstract

Learned access control is typically deployed behind two hand-built artefacts: a hard-deny set enforced by a small trusted shield, and a region map that decides which requests mined rules answer and which the learner answers. Verifying the shield or measuring the learner validates neither. We audit both on the five public ABAC Lab policies. First, we formalise hard-deny soundness and check it exhaustively over relations of up to 794,250 tuples. Clauses that faithfully transcribe an explicit restriction are unsound at rates of 0.50-0.70 across four elicitors - an author and three blind language-model elicitors - because prose scopes restrictions to a section while the grant relation is a disjunction over all rules: 97% of counterexamples are granted by other rules. Shipped unchecked, these deny sets would withhold 16-78% of everything the policies grant; certified, they withhold nothing. Vacuity, a second silent failure, also occurs. Second, hybrid routing needs a two-class, abstaining rule set that ABAC miners do not emit, and ranking a mixed-effect candidate pool under a cap is one-class in all 24 cells for the two commonest ranking keys. Third, for additive losses the hybrid-versus-learner effect factorises into covered mass times a rule-learner risk contrast that must be reported on both false-non-grant and false-permit loss: on two real policies, most selection rules give opposite verdicts on the two losses, and the one favourable on both holds for one policy, one learner configuration and a band of noise levels. A released artefact regenerates and checks every reported number.

0.20

WADE: An LSTM-Based Anomaly Detection Engine for WebAssembly Applications

cs.CR cs.LG ·2026-10-22 ·Ilaria Campestrini, Stefano Zanero, Engin Kirda
Abstract

WebAssembly (Wasm) enables near-native performance for web applications through a portable and sandboxed execution model. However, its limited runtime protection mechanisms (e.g., the lack of stack canaries and address space layout randomization (ASLR)) leave Wasm applications vulnerable to traditional attacks. We present WADE (WebAssembly Anomaly Detection Engine), a system that detects anomalous runtime behaviors in browser-executed WebAssembly applications. WADE monitors function-level execution traces collected under benign conditions and trains a Long Short-Term Memory (LSTM) network to learn the normal execution behavior of each application. At runtime, the model flags deviations from these learned patterns as potential attacks. Unlike signature-based approaches, WADE builds application-specific behavioral models, enabling detection of both known and previously unseen attacks without per-vulnerability signatures. We evaluate WADE on a dataset of 67 WebAssembly applications spanning ten vulnerability classes, achieving zero false negatives and a 0.42% false positive rate.

0.18

The Entire Quantum Computer on a Classical Leash

cs.CR quant-ph ·2026-10-22 ·Yael Tauman Kalai, Dakshita Khurana, Justin Raizes
Abstract

We show that a classical client with no quantum hardware can use a single untrusted and potentially malicious quantum server as if it were its own trusted quantum computer. Through the server, the client can privately choose and run quantum computations, prepare quantum states, and process quantum states supplied by other parties. It can also keep quantum memory on the server, hidden from the server itself, and use that memory in later computations so its quantum state persists across rounds of interaction. Together, these capabilities let a purely classical party act as a quantum party in a variety of interactive protocols. We expect this virtual quantum computer to be a broadly useful tool. As immediate applications, we resolve two known open problems: (1) classical verification of quantum sampling with negligible computational soundness error, and (2) remote state preparation of arbitrary efficiently preparable quantum states. Underlying all of these results is a general-purpose, maliciously secure two-party computation protocol between a classical client and a quantum server, which we construct assuming only the post-quantum hardness of learning with errors.

0.15

Amplifying Latency and Cost of LLM Inference via Rowhammer Attack

cs.CR cs.LG ·2026-10-22 ·Mrityunjay Shukla, Shubham Roy, Biswabandan Panda
Abstract

Large language models (LLMs) achieve high inference efficiency using optimized GPU kernels generated by just-in-time (JIT) compilation frameworks such as Triton. Triton compiles GPU kernels through an MLIR/LLVM-based pipeline. We investigate whether a Rowhammer-induced bit-flip in the CPU's DRAM can degrade GPU-based LLM inference performance while preserving output correctness. We present LLVMhammer, the first Rowhammer attack that causes LLM inference slowdown by inducing a bit-flip in Triton's compilation pipeline. A bit-flip in a performance-critical optimization pass causes Triton to generate a functionally identical but suboptimal GPU kernel. Since Triton caches the generated kernel on disk, a single successful bit-flip causes persistent slowdown across subsequent inference executions. Triton's target-independent stages are shared across GPU vendors, which makes our attack GPU vendor-agnostic. To identify exploitable bit-flip locations without manual inspection, we build Karnage, a tool that isolates useful bit-flip locations in Triton's target-independent compilation pipeline. We evaluate LLVMHammer across a diverse set of LLMs, including Qwen2.5-7B-Instruct, Llama-3.1-8B-Instruct, Mistral-7B-Instruct-v0.3, and Deepseek-llm-7B-chat, across NVIDIA RTX 4090 and AMD RX 7900 XTX GPUs. LLVMhammer increases inference latency of an LLM inference server running on a single GPU by 4-16\%, which, in turn, can translate into additional annual operating costs at the scale of multi-GPU deployments. We further show that LLVMhammer can also corrupt LLM outputs across different GPU vendors, reducing BERTScore to as low as 25\%, illustrating the broader security implications of bit-flips in Triton's compilation pipeline.

0.14

Not an Accident: Exploiting Safety Filter Inadequacy in Humanoid Robots

cs.CR cs.RO ·2026-10-22 ·Tiffany Bao, Shangtao Li, Yifan Sun, Mengwen He, Bowei Li +4 more
Abstract

As humanoid robots move from research laboratories toward deployment in human-centered environments such as homes, hospitals, and warehouses, safe operation around nearby people and objects becomes a fundamental requirement. Safety filters have emerged as a promising technique to address this challenge by providing mathematical guarantees that generated control actions do not result in collisions. This paper shows that these guarantees hold only when the safety parameters are configured correctly. However, correct parameter configurations are task- and environment-specific, and no general method exists to determine them across diverse deployments. As a result, safety filters shipped with humanoid robots can fail after deployment in specific scenarios. We show that attackers can exploit these failures through only normal user interactions without any additional privileges or capabilities. To demonstrate this risk, we present goal insertion attacks and develop SIREN, an automated tool that identifies vulnerable inserting goals for a given task and environment. Experimental results show that SIREN identifies attacks in all vulnerable scenarios in the SPARK dataset. Compared with a baseline approach, SIREN discovers three times as many attacks. We also show that SIREN is able to generate attacks against a real humanoid and cause it to collide while executing tasks. These results demonstrate that current safety filters introduce exploitable vulnerabilities that can cause safety violations, which reveals that existing safety mechanisms can themselves become a source of safety risk for humanoid robots.

0.13

HiMark: Hierarchical and Controllable Multi-bit Watermarking for Large Language Models

cs.CR ·2026-10-22 ·Ke Zhu, Yanli Chen, Ling Xiong, Gejian Zhao, Hanzhou Wu
Abstract · To appear in IEEE International Workshop on Information Forensics and Security (2026)

With the rapid adoption of large language models (LLMs) in text generation applications, it has become increasingly difficult to trace generated content and assign responsibility for its use, raising concerns about provenance, copyright protection, and accountability. Multi-bit watermarking addresses these issues by embedding multi-bit provenance information into generated text, but reliable extraction remains difficult due to sparse per-bit evidence and the trade-off between watermark-extraction accuracy and text quality. To deal with these challenges, we propose HiMark, a hierarchical and controllable multi-bit watermarking framework for LLM-generated text. During watermark embedding, HiMark uses a hierarchical candidate-control mechanism, including initial candidate sampling, layer-wise tournament elimination, support-based refinement, and support-threshold filtering. During watermark detection, a model-agnostic evidence aggregation mechanism computes multi-layer support evidence for each observed token, and then accumulates these token-level votes into the corresponding information for robust watermark extraction. Experiments show that HiMark improves watermark extraction accuracy while maintaining text quality, e.g., the watermark extraction accuracy is 90.01% under the paraphrasing attack when 32 bits were embedded within a text containing 300 tokens, demonstrating its effectiveness for provenance tracking.

0.12

Systems & Networking 25

TT-SAX: A Two-Track Time Series Symbolization for Natural Language Shape Querying

cs.DB cs.CL ·2026-10-22 ·Zhao Tan, Xianglong Bao, Zhipeng Xi, Ming Jin
publishedbenchmarkcode/data
Abstract · 14 pages, 8 figures, 2 tables. Accepted at the Australasian Database Conference (ADC 2026). This is the submitted manuscript prior to peer r…

Natural language querying for time series databases aims to let users find what they need in long histories of data by asking in plain language. Shape queries ask how a series moves. They are the main challenge of this task. Recent work makes them feasible by using symbolization as a bridge between the query and the long history. However, existing symbolizations were designed for compression, similarity indexing, or classification. What a symbolization should provide for natural language querying has not been systematically studied. We therefore examine the definitions of six classical symbolizations against this task and identify three limitations. (1) Direction is missing from the string, (2) direction is measured at too small a scale, and (3) the letters are arbitrary. Three design principles follow. We follow these principles and make a minimal redesign of SAX called TT-SAX (two-track SAX). The level letters stay. Each segment gains a trend letter. It is measured across neighboring segments rather than inside one, and graded to match the adjectives that queries use. We conduct extensive experiments on the public NLQTSBench benchmark, comparing TT-SAX with the six classical symbolizations across six LLMs. TT-SAX outperforms all six on shape queries, and ablations trace its gains back to the three limitations one by one. This paper is the first systematic study of time series symbolization with natural language querying as the design goal. It provides a foundation for future research on this task. Our code is available at https://github.com/Atlamtiz/tt-sax.

0.52

Fast Path, Exact Path: One Library of Hardwired Operators for Streaming and Exact Canny Edge Detection on a Low-Cost FPGA

cs.AR eess.IV ·2026-10-22 ·Raj Mali
Abstract · 7 pages, 5 figures, 7 tables

Hysteresis edge tracking, the final stage of Canny edge detection, follows edge chains of unbounded length and therefore resists single-pass streaming; FPGA designs either approximate it or dedicate separate labelling hardware to it. We show that a single library of hardwired 3x3 operators, each with an identical two-clock latency, delivers both a camera-rate approximation and exact hysteresis on a low-cost Artix-7 (XC7A35T). The fast path streams camera pixels through the operators and K local hysteresis stages and commits each edge decision 262.15-720.63 microseconds after the pixel arrives (K = 1-8; system simulation with a datasheet camera model), that is, 3 + K camera lines later and 46-127x sooner than a frame-based processor can even begin at 30 frames per second. It provably never reports an edge that exact hysteresis rejects, recovers 76.6-96.1% of exact edges at thresholds 20/40, and on the 200 BSDS500 test images retains at least 98.6% of the exact path's boundary F-measure (ODS 0.470-0.477 versus 0.476). The exact path holds 128x128 tiles in block RAM, is exact within each tile, and replaces two-buffer iteration with in-place forward and reverse sweeps built into the same operator. This reduces hysteresis from 38-82 passes to 3-7 on pattern and photo tiles, accelerating the hysteresis step 6.6-21.7x (40x on a worst-case tile) and a complete tile 4.5-9.9x, with identical output. Both paths are bit-exact with software golden models at every pipeline stage and close timing at 50 MHz in at most 1455 LUTs. An optimised multi-core CPU remains faster on stored images, but it cannot begin until the frame is complete.

0.51

Pipelining vs. Combining in Aggregation Tasks

cs.DC cs.DS ·2026-10-22 ·Gabriel Marques Domingues, Claire Mathieu, Boaz Patt-Shamir
Abstract · 22 pages; 7 figures

We study the aggregation task (convergecast) in networks in the case that the values to be aggregated do not fit in a single message. We assume that every node starts with an input value, and there is an aggregation operator that is commutative, associative, and length-preserving, i.e., the result of aggregating two values has the same length as each of the input values. Since values do not fit in a message, they are transferred in a ``train'' of pipelined messages, which must be terminated whenever we want to aggregate values. The goal is to minimize the makespan time of aggregating all values. We present general lower bounds on the problem and optimal algorithms for several prevalent parallel computing architectures, including the hypercube and complete bipartite graphs, under both the splittable and unsplittable flows models. We also present an $O(\log n)$ approximation for general graphs in the unsplittable flows model.

0.40

uGDS: A User-Space Development Kit for High-Performance GPU Direct Storage

cs.DC cs.OS ·2026-10-22 ·Jing Wei, Guanyi Chen, Qi Chen, Chuyu Zhang, Zheng Zhou +2 more
code/data
Abstract · 17 pages,16 figures, Technical report. Source code: https://github.com/ScaleX-IO/uGDS

Data-intensive GPU applications increasingly rely on NVMe SSDs to extend GPU memory. NVIDIA GPU Direct Storage (GDS) enables direct SSD-GPU transfers via PCIe peer-to-peer DMA, eliminating host staging. However, GDS retains three key limitations: (1) its control path traverses the kernel for every I/O request, introducing software overhead; (2) its stream-ordered path imposes strict total ordering even when workloads only require batch-level ordering; and (3) its closed-source user-space engine hinders diagnosis and custom scheduling. We present uGDS, https://github.com/ScaleX-IO/uGDS, an open-source user-space GPU Direct Storage library built on four principles: user-space control, unified cross-vendor abstraction, workload-aware scheduling, and end-to-end transparency. The host CPU manages the entire NVMe command lifecycle in user space, while payloads move via PCIe P2P DMA without consuming GPU compute. A minimal kernel module is used only for one-time privileged setup. uGDS partitions hardware queues between synchronous and batch traffic, enforces DMA-safe buffer lifetimes, and provides cuFile-compatible synchronous, batched, and stream-ordered APIs with polling or interrupt completion on both NVIDIA CUDA and AMD HIP. Evaluated on an A100 with a Samsung 990 PRO, uGDS achieves up to 12.12x GDS's batch-read bandwidth and 30.69x its batch-write bandwidth (8- and 4-KiB requests, batch size 128), and reduces 4-KiB synchronous write latency by 84.1% with two threads. Its stream-ordered path sustains 5.49 GiB/s on a single stream, where GDS retains only 40.8% of its synchronous bandwidth. In LMCache, uGDS reduces median time-to-first-token (TTFT) by up to 62.7% and raises serving goodput by 2.63x. On AMD, uGDS achieves up to 1.19x hipFile's synchronous-write bandwidth.

0.39

RV565: Cross-Layer Co-Design for Frame-Independent Video Playback on MicroPython-Class ESP32-S3 Systems

cs.AR cs.MMcs.PFeess.IV ·2026-10-22 ·Ayushman Bhattacharya, Anwesha Chakraborty
Abstract · 14 pages, 4 figures, 7 tables. Artifact: https://doi.org/10.5281/zenodo.23543367

Video on SPI-attached microcontroller displays is limited less by decoding than by where each byte lives and which stages run serially. We present RV565, a decode framework for frame-independent video whose decoded representation is the panel's native RGB565 format, and evaluate it on an ESP32-S3 with two ten-second windows of an openly licensed film. A serial path - inflate, then scale into internal-memory stripes overlapped with SPI direct-memory-access (DMA) transfers - meets the 41.7 ms deadline on the lighter window but misses all 7,200 frames of the heavier one. RV565's dual-core pipeline inflates the next frame on the second core during the transfer and sustains 24 FPS with zero misses and 10.3 ms of slack on both windows; frame time then equals the display transfer, which runs at 98.1% of the 40 MHz wire-time bound. Because frames are independent, inflation time is linear in the compressed length stored in the container (pooled r^2 = 0.9987): a model fitted on one window predicts the other within 0.04 ms and identifies from metadata alone which window the serial path cannot schedule. We show why Deflate cannot exploit co-located temporal redundancy in RGB565 frames larger than 128x128 pixels and measure the price: a shared Deflate stream changes size by at most 1.1%, whereas an explicit XOR delta would save 46-80%. Inside OreoOS, a MicroPython application operating system, firmware inflation alone reaches 12.2 FPS; stage timing attributes the 80.4 ms frame time to serialized filesystem input, external-RAM placement, and blocking presentation, with 0.8 ms of interpreter overhead. Firmware, measurements, and analysis scripts are released.

0.39

NN-Based Transistor Modeling for Ultra-Fast SPICE Simulation and Standard Cell Library Characterization

cs.ET cs.LG ·2026-10-22 ·Rodion Novkin, Hussam Amrouch
benchmark
Abstract · 11 pages

Neural network (NN)-based transistor models have proven to be a promising solution to accelerate device modeling. Although these models demonstrated remarkable speedup in circuit simulations, they were not rigorously tested in complex tasks within standard EDA tool flows for chip design, which require very high transistor model precision to provide accurate results and converge properly. To investigate their full potential and promise, we develop NN-based transistor models that accurately predict the drain current and various charges for a 3nm nanosheet, and employ the models for a standard cell library characterization. This enables us to evaluate the accuracy and speedup under a very wide range of SPICE simulations, benchmarking our models against an industry-standard implementation. Since standard cell library characterization involves thousands of SPICE simulations, it serves as a great benchmark for transistor model quality and consistency. To further evaluate model accuracy, we also perform analog- and digital-level simulations for various complex circuits. Our experiments reveal that NN-based models are a valid alternative to existing transistor compact models, capable of producing accurate delay and power estimates with a much higher speed and convergence time compared to the conventional Verilog-A-based industry-standard (BSIM-CMG) transistor compact model. Our NNs achieve sub-0.11% and 0.76% errors for delay and total power evaluation on a wide range of circuits, while taking up to 8 times less time during the standard cell library characterization. Additionally, we provide comprehensive information and guidance on NN accuracy, unveiling the relationship between NN size and its effect on circuit simulation precision.

0.35

Energy-Aware Performance Characterization of Kokkos Kernels Across Precision Levels

cs.PF ·2026-10-22 ·Maria Patrou, Andrey Prokopenko, Aditya Kashi, Oscar Hernandez, Jean-Luc Fattebert
publishedbenchmark
Abstract · 13 pages, 4 figures, Accepted at the 3rd International Workshop on Energy Efficiency with Sustainable Performance: Techniques, Tools, and Be…

Kokkos is a performance-portable programming model used in GPU-accelerated HPC applications comprising kernels with diverse computational and memory behaviors. We characterize kernels from two applications to evaluate how FP32, FP64, and mixed FP32/FP64 affect arithmetic intensity, performance throughput, and GPU energy consumption. We introduce an Energy-Aware Roofline Model that incorporates energy into performance analysis and a Thermal-Aware Inter-Kernel Energy Model that visualizes energy consumption alongside duration, GPU temperature, and execution order. We study both applications with: (1) fine-grained, application-level instrumentation through the Kokkos Performance Tools PAPI Connector and (2) coarse-grained PAPI instrumentation applied to a dedicated-kernel benchmark using vendor-specific power management components. Our analysis shows that precision effects vary by application, kernel, and workload. Mixed FP32/FP64 kernels generally achieve arithmetic intensity comparable to FP32, with modestly lower throughput and, in most cases, a modest increase in energy consumption. FP64 kernels generally exhibit the highest energy consumption among the evaluated precision configurations.

0.35

Ray-by-Ray: Fine-Grained Resource Management for GPU Ray Tracing Units

cs.AR ·2026-10-22 ·Junrui Pan, Cesar Avalos Baddouh, Kai Ze Ee, Timothy G. Rogers
top venuecode/data
Abstract · to be published in 59th IEEE/ACM International Symposium on Microarchitecture

Ray tracing (RT) generates photorealistic images by simulating the physical behavior of light as it interacts with surfaces in a virtual scene. In film production and high-end rendering pipelines, this technique computes global illumination, reflections, refractions, and soft shadows by tracing millions of light rays per frame and accurately modeling their bounces and attenuation. Contemporary GPUs incorporate specialized ray tracing hardware to accelerate key operations in the traversal pipeline. By accelerating these stages in hardware, the unit enables real-time ray tracing for complex scenes that previously required offline computation. However, ray tracing introduces substantial control and data divergence. Each ray may traverse a different set of nodes in the acceleration structure, encounter distinct materials, or terminate at varying depths. This irregularity reduces thread coherence and exposes a fundamental disparity between the GPU SIMT and per-thread execution model of dedicated RT units, where each ray progresses independently. In this work, we seek to validate assumptions made in open-source infrastructure and identify potential inefficiencies in resource management within the ray tracing pipeline. To address inefficiency caused by mismatches between SIMT and RT granularities, we propose Ray-by-Ray, which decomposes on-chip buffers into the granularity of individual rays and introduces lightweight management structures that enable more warps to run in parallel without increasing buffer capacity. This approach improves ray tracing unit utilization and increases memory-level parallelism. Experimental results show that Ray-by-Ray achieves an average performance improvement of 1.91x over the baseline, with an area overhead of 0.19\%.

0.34

Almost tight distributed approximation for domination in sparse graph classes

cs.DC cs.DMcs.DS ·2026-10-22 ·Timothé Picavet
Abstract · 27 pages, 4 figures

We study distributed approximation of Minimum Dominating Set in a very general class of sparse graphs, extending near-optimal approximation results from the minor-free setting to bounded expansion. In more detail, for fixed $d$ and $\varepsilon>0$, we give a new deterministic constant-round LOCAL algorithm for Minimum Dominating Set on graphs of maximum average degree at most $d$ and bounded depth-one shallow-minor density. Its approximation ratio is $d+1+\varepsilon$, which is optimal up to $\varepsilon$ for every even integer $d\geq 2$. This improves on the recent results of Amiri and Siebertz by extending beyond proper minor-closed classes. From this algorithm, we derive almost-tight constant-round approximations for a number of graph classes. We get $(7+\varepsilon)$-approximations for planar graphs and graphs of bounded Euler genus, and $(5+\varepsilon)$-approximations for their triangle-free subclasses. This answers the $7+\varepsilon$ question of Wawrzyniak positively, and matches the recent results of Amiri and Siebertz. We also apply our algorithm to classes of bounded queue-number or stack-number, obtaining bounds that are tight up to $\varepsilon$. Additionally, we answer questions of Amiri and Siebertz about improving the dependence of the round complexity on $1/\varepsilon$ and obtaining polynomial local computation for fixed parameters, when comparing the output to the usual domination number.

0.33

Finite Systems in the Space of Projective Measurements

cs.ET quant-ph ·2026-10-22 ·Miguel Flores R., Carlos A. Muñoz
unusual
Abstract · 15 pages, 13 figures, 27 images

We investigate the dynamics of two-level systems using the quasidistribution $Q$ function projected onto the space of collective measurements. Particular attention is given to evolution under Hamiltonians that are invariant under particle permutations, for pure states that are either invariant or noninvariant under such permutations. This requires constructing an approximate analog of the star product between quasidistribution functions: the projected $Q$ function of a product of two operators is expressed approximately in terms of the individual operators' $Q$ and $P$ functions. We rewrite the Liouville equation to determine approximate state dynamics and compare the result with exact evolution using the Hellinger distance. Overall, the method performs well for permutation-invariant pure states and Hamiltonians, whereas deviations arise for arbitrary pure states and noninvariant Hamiltonians.

0.32

ChainSEED: A Scalable, High-Fidelity Blockchain Emulator

cs.NI ·2026-10-22 ·Qishu Jin, Wenliang Du, Kairui Feng
Abstract · 19 pages (including references and appendices), 12 figures, 11 tables

Researchers need blockchain experimental environments with configurable workloads, controllable network conditions, and observable outcomes. Public blockchains and testnets limit experimental control, model-based simulators omit real client implementations, and real-client environments face challenges in combining network control with resource-efficient deployment. We present ChainSEED, a scalable, high-fidelity blockchain emulator built on the SEED Internet Emulator. It separates blockchain configuration from client placement on a configurable routed network. For Ethereum PoS, it organizes separately configured execution, consensus, and validator clients and provides APIs for customizing genesis state, peer connections, and client software. A transaction generator informed by Ethereum mainnet activity, application interfaces, and chain- and node-level monitoring support experiments. ChainSEED also supports Monero and Solana. We examine beacon-node memory reduction, execution-client memory allocation, and validator-client multiplexing to support larger experiments within limited hardware budgets. On a single server with approximately 1 TiB of memory and 384 logical CPU cores, ChainSEED deploys 400 beacon nodes, 400 execution nodes, and 60,000 validator identities while sustaining approximately 20 TPS. It also emulates a 1,200-node Monero network with up to 200 active single-threaded PoW miners and a 200-node Solana network with sustained finalization. Case studies demonstrate routing attacks, blockchain amplification attacks, Monero eclipse attacks, topology-discovery benchmarking against known peer graphs, and network-configuration experiments.

0.31

SnakeCharmer: Orchestrating Utility, Timeliness, and Placement in GPU Prefetching

cs.AR ·2026-10-22 ·Yuhan Tang, Tiejun Li, Jianmin Zhang, An Hu, Guangyu Sun +4 more
top venue
Abstract · 15 pages, 26 figures, 3 tables. Accepted at the 59th IEEE/ACM International Symposium on Microarchitecture (MICRO 2026)

GPUs rely on massive thread-level parallelism to tolerate long memory latency, yet they still suffer severe stalls when many warps simultaneously wait for data from the memory system. Address-prediction-based GPU prefetchers can mitigate such stalls, but accurate prediction alone does not guarantee performance improvement: a predicted prefetch becomes beneficial only if it is worth issuing, arrives at the right time, and is retained without harmful interference. Because these requirements interact, optimizing any single dimension in isolation can simply move the bottleneck to another. We capture these requirements with a predicted-prefetch-to-realized-benefit perspective and propose SnakeCharmer, a unified backend prefetch-control framework for address-prediction-based GPU prefetchers. SnakeCharmer separates address prediction from prefetch realization: the frontend predictor generates prefetch candidates, while the backend control path jointly controls utility, timeliness, and placement by determining whether each candidate should be issued, when it should be issued, and where the returned data should be retained. SnakeCharmer realizes this control path with zero-value-aware filtering for utility, hierarchical feedback-driven distance tuning for timeliness, and a register-assisted prefetch buffer for placement. We design and implement SnakeCharmer primarily on top of Snake, and adapt the same backend-control principles to APOGEE to demonstrate that SnakeCharmer's benefits are not limited to a single frontend predictor. On an A100 GPU model, SnakeCharmer improves performance/energy efficiency by 12.8%/20.8% over Snake and by 1.7%/3.2% over APOGEE, with modest hardware overhead.

0.31

Controlled Radio Fault Injection and Telemetry Evaluation for Open 5G RANs

cs.NI eess.SY ·2026-10-22 ·Babak Azkaei, Maxim Bîtiu, Georgios Exarchakos, Kishor Chandra Joshi
published
Abstract

Radio access network (RAN) telemetry reveals radio conditions, protocol behavior, and their consequences for applications. Evaluating its usefulness requires controlled impairments and observations of the same execution. We present C and Python channel brokers and a fault injection workflow for software 5G testbeds connecting an OCUDU base station and a software UE through ZeroMQ. The brokers apply scheduled, directional transformations to I/Q samples and restore the link, while the workflow jointly records applied faults, native and programmable (codelet) telemetry, and application delay. We demonstrate this capability through 45 comparative trials of brief uplink interruptions and stepped additive noise. These use cases distinguish recognizing a radio change from measuring its service consequence: native missing-feedback events retrospectively identify interruptions, whereas noise changes decoding-error measurements without producing those events. Median on-time round-trip times remain near 30 ms, but the prespecified comparisons show mean increases in on-time 95th-percentile delay of 354 and 604 ms for one and five interruptions, and 113 ms for noise. Adding receive-side radio link control (RLC) codelet summaries did not improve interruption classification on held-out trials. The framework provides a reusable way to evaluate measurement choices under known radio faults, before using them in monitoring or learning systems.

0.31

EGGPlacer: Energy-Guided Generative Chip Placement with Diffusion Models

cs.AR cs.AI ·2026-10-22 ·Qi Zhou, Guojun Liu, Juanhong Zhao, Guangzhi Qi, Zehao Chen +1 more
Abstract · 27 pages, 9 figures, 11 tables. Preprint

Chip placement strongly affects routed wirelength and thermal behavior, yet these downstream metrics are difficult to evaluate during macro placement. Learning-based macro placers commonly optimize the macro-only half-perimeter wirelength (HPWL), denoted mHPWL, which ignores standard cell connectivity and spatial demand. In contrast, analytical mixed-size placers optimize differentiable full netlist wirelength and density objectives but are typically unaware of thermal effects. We propose EGGPlacer, an energy-guided diffusion framework that incorporates analytical placement feedback and thermal feedback into macro placement generation. An analytical placer acts as an online physical response model, exposing the standard cell distribution and a full netlist wirelength proxy induced by each intermediate macro placement, while a differentiable thermal model provides temperature feedback. These signals are formulated as composable energies and injected throughout reverse diffusion without retraining the underlying generative model. Experiments on public placement benchmarks show that EGGPlacer achieves competitive mixed-size HPWL relative to strong analytical and learning-based macro placers. In the end-to-end evaluation, EGGPlacer also obtains the lowest average routed wirelength and peak temperature among the evaluated methods. Ablation studies further demonstrate that analytical and thermal guidance provide distinct and complementary benefits across wirelength proxies, routed wirelength, and thermal objectives.

0.29

Low-Latency MoE Inference in Space Data Centers: A Wireless-Power-Transfer Perspective

cs.DC cs.AIcs.NI ·2026-10-22 ·Haotian Zheng, Zhanwei Wang, Kaibin Huang
Abstract · 13 pages, 11 figures

Space data centers (SpaceDCs) are emerging as platforms for large language models (LLMs), supported by continuous solar-energy harvesting and high-speed interconnections. The mixture-of-experts (MoE) architecture supports distributed LLM inference by deploying experts across satellites. However, skewed expert activation creates uneven computation power demands despite uniform solar-power supply at the dawn-dusk sun-synchronous orbit (SSO). Consequently, active experts remain power-constrained while harvested power at idle experts is underused. These reduce inference efficiency and increase latency. To address this supply-demand mismatch, we propose the SpaceWPT framework, which uses inter-satellite laser-based wireless power transfer (WPT) to redistribute harvested power. The key design challenge is to balance the computation demands of power-donating and power-receiving satellites under WPT losses: transferring more power accelerates computation at the receiver but leaves less power for the donor. After deriving design principles that simplify WPT control, we characterize the resulting latency tradeoff through a tractable surrogate of average inference latency. Minimizing the surrogate yields a novel two-sided water-filling strategy that determines whether each active expert donates, retains, or receives power according to the water level and its WPT link efficiency. Simulations with Mixtral-8x7B and Phi-3.5-MoE show up to 29.25% lower inference latency than without WPT, with larger gains under more compact formations and higher transfer efficiencies.

0.29

Repair Changes Which Layers to Remove: Exact Optima for Depth Pruning of Language Models

cs.DC ·2026-10-22 ·Jianchang Su, Dawei Xiang, Yifan Zhang, Wei Zhang
hot direction
Abstract

Depth pruning removes entire transformer layers to make large language models faster, and recent methods recover quality by replacing the removed layers with a training-free linear map. Which layers to remove, however, is chosen with heuristic scores whose distance from the best choice is unknown. We evaluate every subset of layer groups exactly (about 53,000 pruned models across four families, with and without repair) and find that the best layers to remove depend on the repair: linear repair absorbs the interaction among nearby layers and turns the optimum from a split layout into a contiguous block, so methods that choose layers without the repair miss the repaired optimum by up to 189%. We therefore select layers under the repair that will be applied. A greedy search at the cost of SLEB lowers perplexity by 20% to 66% relative to ReplaceMe on five models up to 27B parameters, and with a task objective it is up to 25 points more accurate given the same data. A second-order expansion of the loss yields a score that selects the repaired block from one forward and one backward pass of the intact model.

0.29

OMRC: Multipath Reliability for Direct-Connect Optical Interconnects

cs.NI ·2026-10-22 ·Eric Ding, Keshav Pradeep, Rachee Singh
Abstract

Network failures degrade Multipath Reliable Connection (MRC) in direct-connect optical datacenters. We propose Optical MRC (OMRC), which remaps collectives and circuits, preserving 95.7% of training throughput, restoring stalled inference, and achieving 4.7x MRC's simulated bandwidth.

0.27

HDLouvain: High-Throughput Incremental Clustering for Hyperdimensional Data

cs.AR ·2026-10-22 ·Beenish Gul, Mahbod Afarin, Farhat Ahmed, Sumukh Pinge, Tsung-Han Lu +2 more
benchmark
Abstract

Modern shotgun proteomics generates tens of thousands of tandem mass spectra per hour, producing hundreds of gigabytes of data that strain conventional clustering pipelines. DBSCAN and hierarchical agglomerative clustering (HAC) offer flexible similarity search, but DBSCAN is unstable across varying data densities, HAC costs $O(n^2)$, and both yield poor cluster quality. Graph-based methods such as Louvain produce high-quality clusters but scale poorly on streaming data. We observe that Louvain's poor scalability can be resolved without sacrificing its quality, and present HDLouvain, which combines Louvain community detection with the bit-parallel similarity computation of hyperdimensional computing (HDC). HDLouvain partitions a bootstrap fraction of the stream into precursor-mass buckets, applies Louvain to each bucket's kNN similarity graph, and condenses each community into a single binary hypervector by bitwise-majority bundling rather than iterative centroid recomputation. New spectra are assigned by a GPU-accelerated Hamming distance lookup against representatives in their mass bucket, or seed a new cluster when no match meets a similarity threshold $τ$, so the partition grows incrementally without full re-clustering. On five PRIDE datasets (0.99M to 21.13M spectra), sweeping the bootstrap fraction $α$ from 0.2 to 0.9 exposes an explicit, monotonic throughput-fidelity trade-off. On PXD003258, speedup ranges from 4.5x at $α=0.3$ to 1.65x at $α=0.7$. Modularity rises from 0.75-0.76 at $α=0.2$ to 0.95-0.97 at $α=0.9$, never more than 8% below full static recomputation. The Louvain stage achieves 0.86-0.93 normalized mutual information (NMI) on LFR benchmark networks with 30k to 2M nodes, whereas DBSCAN and HAC degrade to 0.57.

0.26

AERIAL: Mission-Aware LLM Selection for Language-Driven UAV Control in Flying Networks

cs.NI ·2026-10-22 ·Diana Nunes, Ricardo Amorim, Pedro Ribeiro, André Coelho, Rui Campos
benchmark
Abstract · 13 pages, 3 figures

Unmanned Aerial Vehicles (UAVs) in Flying Networks (FNs) increasingly require mechanisms to translate high-level natural language objectives into executable missions. Although Large Language Models (LLMs) have demonstrated the feasibility of language-driven UAV control, existing approaches typically rely on static model configurations and overlook mission-aware performance variability. This paper presents AERIAL, a benchmark-driven framework for adaptive LLM selection in hybrid edge-cloud UAV control. AERIAL separates systematic model benchmarking from mission execution, deriving empirical scenario-model mappings and applying them at runtime for mission-aware selection. Evaluation across 720 automated simulation runs spanning navigation, spatial reasoning, and algorithm invocation revealed significant performance variability depending on scenario-model pairing. A case study integrating AERIAL with a FNs placement algorithm further validated the framework's ability to interface with domain-specific optimization modules through natural language. These results validate the effectiveness of AERIAL's adaptive model selection in improving reliability and operational robustness for autonomous UAV control in FNs.

0.25

Aegis OS: A Transactional Multi-Tenant Operating System Runtime for Eliminating Authorization Fatigue in Autonomous Coding Agents

cs.OS cs.CRcs.MAcs.SE ·2026-10-22 ·Hui Zhao
code/data
Abstract · 19 pages, 3 TikZ figures, 13 tables. Open-source Rust implementation and reproducible benchmark suite available at https://github.com/heurio…

Autonomous coding agents such as Google Antigravity (agy), Anthropic Claude Code (claude), and OpenAI Codex CLI (codex) execute multi-step software engineering workflows on workstations and cloud hosts. Existing agent environments face a Security-Autonomy Dilemma: operators must choose between per-action confirmation prompts, which interrupt workflows on 45%-57% of tool calls and induce authorization fatigue, or unconfined execution (--yolo / --dangerously-skip-permissions), which exposes host files, SSH/cloud credentials, and API tokens to prompt injection and hallucination risks. Furthermore, concurrent agents editing a shared working tree clobber one another's changes, while passing large artifacts inline through LLM context windows incurs high token overhead. We present Aegis OS (aegis-os), a transactional, multi-tenant Agent Operating System Runtime in Rust exposed via the Model Context Protocol (MCP) and Google Agent2Agent (A2A) protocol. Aegis OS shifts the authorization boundary from individual tool calls to transactional workspace boundaries. Each agent runs inside a kernel-enforced OS sandbox (macOS Seatbelt and Linux Bubblewrap with default-deny --tmpfs $HOME isolation and a Unix-socket egress bridge) backed by a Copy-on-Write (CoW) shadow workspace (APFS clonefile; single-open Linux FICLONE reflinks), an identity-pinned HTTP Egress Proxy with L7 method/path-scoped secret injection and outbound DLP, and an A2A IPC bus with Content-Addressable Storage (CAS). Across 1,500 controlled calls and 28,087 real calls across 52 coding-agent sessions, Aegis OS reduces prompts by 94.19%-99.88% (controlled) and 90.7%-97.9% (real sessions, 1.12% IPR), lowers attack success rate across 80 attacks in eight categories from 100.0% to 0.0% on macOS and Linux, clones 1,000-file (78.1 MB) repos in 6.3 ms on APFS and 15.6 ms on btrfs, and catches 10/10 syntactic and semantic merge conflicts.

0.22

Workload-Aware Monitoring for AI-RAN

cs.NI eess.SY ·2026-10-22 ·Babak Azkaei, Georgios Exarchakos, Kishor Chandra Joshi
published
Abstract

AI-RAN platforms are expected to share CPU and GPU resources between radio workloads and AI services. Aggregate telemetry reports operating conditions, but timing changes may reflect different workloads, processors, or measurement boundaries. We propose workload-aware monitoring that associates radio-work descriptors with processing paths, timing boundaries, and operating context before comparing like with like. In 16 over-the-air downlink experiments, hybrid dispatch (CPU for transport blocks (TBs) below 4,096 bytes and GPU otherwise) has 29.66% and 22.93% lower mean host-callback durations than direct GPU dispatch, without and with a synthetic GPU co-runner, respectively. Grouping locates the separation in small TBs, while large-TB callback-duration ranges overlap. These observations identify a candidate for complementary CPU/GPU use.

0.21

System-Level Evaluation of Uplink Access Schemes for Ambient IoT

cs.NI ·2026-10-22 ·Douglas Dziedzorm Agbeve, Henrique Duarte Moura, Andrey Belogaev, Jeroen Famaey
Abstract

Ambient Internet of Things (A-IoT) devices operate under stringent energy constraints, relying on harvested power in the $μ$W-to-mW range and limited energy storage. The A-IoT air interface for active devices is still under development; we therefore use Narrowband Internet of Things (NB-IoT) as an implementable reference for evaluating autonomous Uplink (UL) access under harvested energy. Under these conditions, the standard NB-IoT contention-based Random Access (RA) procedure incurs substantial per-packet overhead, making UL access in dense A-IoT networks increasingly energy-expensive. Context-retaining UL access schemes reduce this overhead, but their energy-latency trade-offs are insufficiently studied in literature under realistic harvesting and protocol constraints. To address this gap, we extend the LENA-NB ns-3 module with solar energy harvesting, supercapacitor storage, and a hysteretic brown-out mechanism, and propose Hybrid Scheduling Request (HSR), a hybrid UL access scheme combining reserved request occasions with opportunistic contention. We compare RA, Scheduling Request (SR), HSR and an Idealized Fast Uplink Grant (FUG) serving as a performance bound. Results show that HSR reduces mean packet delay by up to 69% and SR reduces energy per delivered bit by up to 24%, both relative to RA.

0.18

Breaking the Coupling: Efficient Concurrency Fuzzing for Linux Kernel Data Race Detection

cs.OS ·2026-10-22 ·Cen Zhang, Jia-Ju Bai, Zi-Yu Zhang, Zuming Jiang
Abstract

Detecting data races in the Linux kernel requires exploring both the input space and the thread-interleaving space. Existing kernel concurrency fuzzers couple these two searches by exploring interleavings during fuzzing, often spending substantial scheduling effort on inputs with little potential to trigger real races. In this paper, we present a new concurrency fuzzing paradigm that decouples concurrent-input generation from thread-interleaving exploration, enabling expensive scheduling to focus on promising inputs. To identify inputs that warrant targeted scheduling, we introduce May Race Pairs (MRPs), which represent conflicting memory-access pairs observed within a dynamically determined time threshold. Our framework first generates and executes concurrent syscall groups without controlled scheduling, collecting MRPs to guide further input mutation and identify schedule-worthy inputs. It then uses a separate scheduling instance to explore targeted interleavings and confirm real data races. To improve the effectiveness of both stages, we incorporate LLM-assisted semantic mutation for generating diverse syscall groups, backpressure-guided dynamic threshold control for balancing candidate production and scheduling consumption, and history-aware context recovery, context-aware interleaving scheduling, and candidate MRP prioritization for efficient race confirmation. We implement our framework for the Linux kernel and evaluate it on eight common kernel modules in Linux 6.17. It discovers 187 unique real-world data races, including 59 harmful races. Under the same testing budget, our approach detects substantially more data races than existing kernel concurrency fuzzers, including SegFuzz and CONZZER. These results demonstrate the effectiveness of decoupling concurrent-input generation from interleaving exploration and concentrating scheduling effort on promising candidates.

0.12

SARIA: Persistence-Aware KV Caching with Coordinated DRAM--SSD Reload for LLM Agents

cs.OS ·2026-10-22 ·Hui Li, Qihui Zhou, Xiao Yan, Shumeng Zhang, Zijie Dai +5 more
Abstract · 12 pages, 16 figures

Multi-turn LLM agents usually reuse long context prefixes, making key value (KV) caching crucial for reducing prefill latency. SSDs provide a low-cost extension to limited DRAM capacity but introduce two key challenges. Eagerly persisting all KV cache blocks incurs substantial write traffic and rapidly consumes finite SSD endurance. Meanwhile, the small number of SSDs provisioned in typical GPU servers limits aggregate reload bandwidth, making SSD-resident KV cache much slower to retrieve than that in DRAM and inflating prefill latency. We present \textsc{Saria} (Selective Admission and Reload for Inference Acceleration), a KV caching system that jointly manages DRAM and SSD by exploiting the session-level temporal locality of LLM agents to reduce SSD writes and accelerate cache reload. On the write path, \textsc{Saria} uses eviction and reuse history to selectively persist KV cache blocks that exhibit reuse opportunities beyond their DRAM residency, avoiding unnecessary writes for KV cache already hosted in DRAM. On the read path, \textsc{Saria} promotes selected SSD-hit KV cache blocks from GPU memory into DRAM after prefill, allowing subsequent reuse within the same session to reduce SSD read traffic while limiting DRAM contention. A transfer coordinator paces DRAM reads while SSD transfers are active, allowing both paths to make progress and efficiently share GPU ingress bandwidth. We implement \textsc{Saria} on LMCache MP and evaluate it using workloads from TraceLab and WEKA, covering both full-attention and hybrid-attention models. The results show that \textsc{Saria} reduces SSD writes by 57.1--58.9\% compared with popular full duplicate caching and the mean time to first token (TTFT) by 42.3--79.7\% over DRAM-only caching.

0.08

KVStream: Adaptive Parallel Storage Management for KV Caching in Large-Scale LLM Serving

cs.DC ·2026-10-22 ·Avinash Maurya, Jaiaid Mobin, Jie Ye, M. Mustafa Rafique, Franck Cappello +1 more
Abstract

Long-context LLM serving for agentic, coding, and document workloads reuses prefixes whose key-value (KV) cache outgrows GPU and host memory, putting storage I/O on the critical path of the inference step for retrieving prefixes. A single node-local NVMe absorbs all retrievals, while the parallel file system or cloud file store that nodes already mount carries almost no inference traffic. We cast prefix retrieval as aggregate-bandwidth scheduling across these tiers and present KVStream, a KV cache engine that batches a step's reads, defers writes behind the forward pass, replicates each chunk on every tier, and lets each tier adaptively claim the next unread chunks once it is free. Our performance model shows this routing attains the aggregate-bandwidth bound without prediction or tuning, whereas any static split misses it by a margin that grows as per-tier bandwidths drift or fluctuate. It also predicts that the time-to-first-token (TTFT) gain grows with load, as faster retrieval drains the queue, and disappears when host memory or arrival order keeps reuse in cache. Plugged into LMCache and evaluated on four models, six datasets, and up to 128 GPUs, KVStream retrieves within 5% of the bound, delivers up to 1.49$\times$ higher offline throughput than LMCache, and lowers mean TTFT by up to 6.1$\times$ under bursty arrivals.

0.06

Software & Programming 35

Stale Instruction Files Override the Code in Coding Agents: A One-Sentence Fix

cs.SE ·2026-10-22 ·Xiaonan Xu, Wenjing Wu
code/data
Abstract · 14 pages, 8 figures. Submitted to SN Computer Science. Data and code: https://github.com/WenJing95/stale-docs-data

Purpose: Coding agents follow a project's instruction file, such as CLAUDE.md or AGENTS.md, even when it disagrees with the existing code and even when they notice the disagreement. We measure this behaviour and test whether one sentence in the user's own instruction file makes agents follow the code instead. Methods: We built 40 Python projects in which the instruction file asks for one convention while the existing code uses another, for example print instead of logging. Claude Code and Codex CLI added a feature to each project, with and without a sentence in the user-level instruction file that gives the existing code precedence, in 320 runs. Results: Without the sentence, 90.1% of the agents' choices followed the stale file. Claude Code pointed out the conflict in 79 of 80 runs and still followed the file. Codex CLI rewrote the existing code to match the file in 20 of 80 runs, and the project's own tests then failed in 17 of them. The sentence raised the share of choices that follow the code from 4.2% to 91.4%, stopped the rewrites and raised the test pass rate from 89.4% to 100%. Conclusion: We attribute the behaviour to how agents receive the two sources: the instruction file reaches the model as an instruction and the code as tool output, which models are trained to rank lower. Users can add the sentence today, and makers of agents and models can build the same rule into their products.

0.68

ProductWebBench: Evaluating Website Continuity Beyond Task Completion

cs.SE cs.AI ·2026-10-22 ·Jiajun Wu, Jian Yang, Yaxin Du, Shukai Liu, Linzheng Chai +7 more
benchmark
Abstract · 35 pages, including appendix

Frontend coding agents are usually graded on one question: did the requested target get built. Real engineering changes carry a second obligation, because the rest of the site has to keep working. We call that second obligation website continuity, and introduce ProductWebBench, a benchmark that scores both on living frontend repositories: 320 Change tasks over existing code and 80 long-horizon Build tasks. Every task freezes a repository snapshot, staged requirements, browser evidence and deterministic checks before evaluation, and enters only once a reference passes every check and plausible-wrong ones fail. A simulated user reveals the requirements one stage at a time, and every check cleared earlier stays armed for the rest of the run. Across 8,737 runs from 13 models the two come apart: of the Change runs passing every requested-work check, 28.2% still fail a continuity check -- a blank route, horizontal overflow, console errors -- and 10.8% push a browser state they had already passed back to failing. Because the same checks also run on the untouched repository, each regression carries exact provenance, not a post-hoc label: on the 586 runs where both readings exist, 18.3% are caught breaking a verified state. Interaction is a further blind spot, failing in 29.5% of runs whose static DOM already passes. ProductWebBench therefore scores a patch against the living system it lands in, not the target alone.

0.57

Automating Feature Flag Cleanup at Scale with Multi-Agent LLM Systems: An Industrial Experience at DoorDash

cs.SE cs.AIcs.MA ·2026-10-22 ·Atharv Chandratre, Jai Datar
top venuehot direction
Abstract · 6 pages, 2 figures. Accepted at the 2026 IEEE International Conference on Software Maintenance and Evolution (ICSME), Industry Track

Feature flags are runtime configuration mechanisms used for gradual feature rollouts, experimentation, and kill switches. Once a feature is fully rolled out, the flag and its gated code become technical debt and can introduce operational risk through inadvertently enabling dormant code paths. At DoorDash, over 56,000 feature flags span 623 repositories, with 1,659 currently stale as of May 2026, with internal estimates of one to two hours per manual cleanup. Existing rule-based cleanup tools proved insufficient in DoorDash's codebases. Our contribution is an end-to-end automated LLM-based feature flag cleanup system that integrates querying live feature flag rollout data using MCP, Jira ticket processing, and end-to-end pull request creation. Our system successfully produced engineer-approved flag removal PRs for 45 (90%) of the 50 stale flags it was evaluated on, proving the system's operational viability at scale with each cleanup averaging only $4.79 and 13.8 minutes. We evaluated the system's quality and reliability boundary based on success rates, cost per cleanup and time metrics. Furthermore, we discuss our lessons learned when architecting this system, to provide practical guidance for developing similar multi-agent systems.

0.53

ToolContractGate: Versioned Trusted Contracts for Drift-Safe Agent Tool Invocation

cs.SE cs.CR ·2026-10-22 ·Sachin Gupta
publishedbenchmarkcode/data
Abstract · Accepted for publication at the 2026 International Conference on Computer and Applications (ICCA). Author-accepted manuscript; 7 pages, 3 fi…

Agent tool approval and input validation do not, by themselves, bind a request to an approved source artifact. ToolContractGate provides a deterministic runtime contract evaluator for OpenAPI-derived agent tools. Each versioned ToolContract combines a stored OpenAPI fingerprint with access, schema, and behavioral declarations. The full evaluator applies ten checks, including a cached boot-time provenance result; four execution modes and five evaluation configurations support baseline, ablation, and non-blocking shadow analysis. On a synthetic benchmark of 45 cases (35 unsafe cases across 12 drift classes and 10 safe cases), the full pipeline detects all 35 faults: 29 BLOCK, 3 REVIEW, and 3 WARN decisions, with no false blocks on the safe cases. The archived run reports a mean gate latency of 0.18 ms and a 95th percentile (p95) of 2 ms using integer-millisecond measurements. The open-source artifact includes seven OpenAPI specifications, twelve enterprise contracts, 29 unit tests, and a 900-record evaluation. Results characterize contract-decision coverage; downstream enforcement requires a host that honors the returned decision. Production traffic, semantic backend drift, and dynamic provenance refresh remain evaluation and integration opportunities.

0.45

Beyond Accuracy: Disentangling Decomposition from Typed-State Quality Assurance in Multi-Agent LLM Pipelines

cs.SE cs.AIcs.ETcs.LG ·2026-10-22 ·Maria Gabriela Alves Zuppardo, Vinicius Cardoso Garcia
published
Abstract · 10 pages, 10 tables. To be published at Brazilian Symposium on Software Quality -- Simpósio Brasileiro de Qualidade de Software (SBQS)

Multi-agent LLM pipelines are often credited with gains in software engineering tasks, but it is rarely clear whether a gain stems from the typed contract between agents or from the specialized prompts that decomposition enables. We study this attribution problem in Requirements Engineering, comparing three architectures for classifying requirements as functional or non-functional and then prioritizing them with MoSCoW: a single-call baseline; a two-call variant that reuses the pipeline's specialized prompts but passes an unvalidated intermediate; and a pipeline with a Pydantic-validated typed state. The 18 conditions cover PROMISE NFR+ (625 requirements), three locally hosted small language models, and two languages. The F1 gain over the baseline is associated with the two-call, specialized-prompt variant (up to +0.075), whereas the typed state adds a small difference (-0.003 to +0.013) that is significant in one of six comparisons. Beyond accuracy, the typed pipeline contains out-of-range model outputs at the agent boundary, while the unvalidated variant fails later and drops the requirement, and it shows a small, reproducible increase in inter-model MoSCoW consistency whose cause our design cannot isolate. The study is associative: the contrasts bundle prompts, call count, and staged context, and the results come from three 7--8B models on one dataset. The implementation and replication package are archived on Zenodo (https://doi.org/10.5281/zenodo.23380444).

0.45

VERA: A Responsible AI Dashboard for Assessing LLM Deployments in Light of the EU AI Act

cs.SE cs.AIcs.CY ·2026-10-22 ·Robin Quériaux, Solveig Bachellery, Léa Déléris, Saurabhsingh Rajput, Mariam Barry +2 more
publishedbenchmarkcode/data
Abstract · Accepted at the 33rd Asia-Pacific Software Engineering Conference (APSEC 2026), Software Engineering in Practice track, Bali, Indonesia, 7-1…

Institutions that deploy large language models in regulated settings often need technical evidence of responsible AI practices before approving a deployment. COMPL-AI provides a technical interpretation of the EU AI Act that organises this evidence under six principle groups and eighteen technical requirements, twelve of which current benchmarks can measure. The benchmark suites that produce this evidence are typically operated from a command line, while the staff who approve deployments are non-engineers. We present VERA, an open-source self-hosted dashboard deployed to facilitate responsible AI processes within a large European bank. A reviewer selects a model and runs public benchmarks against it. VERA scores the twelve requirements that current benchmarks can measure and instruments the energy consumed by each evaluation, adding run-level evidence for environmental impact. We run four served open-weight models through the requirement pipeline, and the dashboard displays a score for each measurable requirement. A within-subject study then compares how participants read these benchmark results: from the raw output files in one condition, and through the VERA dashboard in the other. The dashboard reduced the time needed to perform the responsible AI assessment, reducing the time required from 56 seconds using raw files to 30 seconds using our dashboard, on average. Our dashboard scored highly (4.36 out of 5) on usefulness in the survey conducted within the organization. The dashboard and the study material are released publicly to promote replicability and reproducibility in different organisational contexts.

0.43

REGATTA: A Cross-Domain Measurement Framework and Fault-Injection Harness for Enterprise AI Agents

cs.SE ·2026-10-22 ·Luong Tuan Thanh, Abhijit Sanyal
publishedcode/data
Abstract · Accepted for publication at 2026 IEEE ICoDSE. 6 pages, 2 figures, 4 tables. Shares the historical experimental corpus and two-stage evaluati…

Enterprise AI-agent evaluation needs evidence about whether test suites expose specified faults and whether their reference rules are trustworthy. We introduce REGATTA (Regulatory Evaluation of Agents via Two-stage Testing and Assurance), a measurement framework with three instruments. A 125-item source-cited checklist spans five industry-by-jurisdiction cells across the United States and Vietnam. A 25-fault corpus covers five categories, and a harness scores suite-level fault detection at the design and execution stages. We examine four measurement diagnostics on 180 generated suites (5,400 scenarios) from three LLM generators. A source audit finds 24 items misstating the rule in force, 46 additional citation defects, 31 items with only non-binding, unidentifiable, or unverifiable sources, and 24 verified clean. Only initially substance-flagged items received author review; all 24 clean rows received AI review only. Results therefore describe the instruments as written, not legal compliance. Item coverage varies, execution-stage fault profiles are stable across generators, and the historical stages agree on $66$--$69\%$ of fault--suite pairs. A current-model US-insurance retest finds stage disagreement above repeated-execution variation, but stage agreement $.536$ falls below pooled chance $.549$, failing V3's lower inequality. Blind adjudication by one expert supports the design-stage judging protocol ($κ=.72$, 95% interval $[.52,.88]$); 40 current-model execution probes give $κ=.75$. The instruments, audit, and reusable code are released.

0.43

Global Convergence of Detailed Balanced Mass Action Systems

cs.LO math.DSq-bio.MN ·2026-10-22 ·Haoran Lv, Chengjin Ni, Manqin Wang, Shenli He
theorycode/dataunusual
Abstract · 23 pages, 1 figure. Lean 4 formalization: https://github.com/BitMiracle-AI/detailed-balance-gac (tag v1.0). Use of AI tools is described in …

We give a self-contained proof that, for a finite detailed balanced mass-action system with positive rate constants, each positive stoichiometric compatibility class contains exactly one positive equilibrium and every positive trajectory converges to it, together with a Lean 4 formalization of the complete argument from persistence to convergence. Boundedness and persistence come from the barrier argument of OpenAI (2026) for weakly reversible systems: an arbitrary-tolerance affine approximation yields a finite-minimum barrier whose superlevel sets confine trajectories and give one logarithmic bound for all solutions from a given initial state. The Horn-Jackson dissipation identity identifies the positive equilibria as an exponential family with parameters in the orthogonal complement of the stoichiometric subspace, which gives uniqueness in each class. A bounded-speed descent argument replaces LaSalle's invariance principle; it uses neither uniqueness of solutions nor continuous dependence on initial data, and it places every cluster point in that family. The route given here is direct and elementary, and its main theorem is checked by the Lean kernel against explicit definitions. The proofs, the Lean development, and the manuscript were produced in ClawsGO Science, an agent harness that packages a complete research environment: Claude Opus 5.5 made the research decisions, GPT 6 Astra carried them out, and the Lean kernel checked every formal step.

0.42

What Ethereum Developers Ask For, and Whether They Get It: Ten Years of Ethereum Stack Exchange

cs.SE ·2026-10-22 ·Tamer Abdelaziz
published
Abstract · 12 pages, Mining Software Repositories

The Ethereum community learns where developers struggle through surveys and interviews, which record what developers say when asked about a chosen theme. Ethereum Stack Exchange (ESE) records what they ask for unprompted, and whether anyone resolves it. We study every question posted to ESE from Jan 2016 to Dec 2025, 54,637 in all, and label each with a prompt-based topic pipeline seeded with eight Ethereum-specific topics and open to new ones (22 topics in all). A three-provider committee of language models screens the labels, and manual adjudication of every disputed case puts their accuracy at 87.9 to 93.5%. Because the corpus is the complete record, no rate carries sampling error. We ask what developers ask for help with, whether the venue still resolves it, and, where it does not, whether the questions changed or their resolution did. Demand is concentrated on writing, deploying, and debugging contracts; the toolchain is the second-largest topic and growing, while security, consensus, and scaling are thin. Supply has weakened, and unevenly: answers still arrive, closure falls further. Across five estimators of the decade, the share of questions drawing any answer fell by 9.5 to 18.6 percentage points, but among answered questions the share reaching an accepted answer fell by 16.2 to 24.5. Rebuilding the later acceptance rate at the earlier topic mix moves the fall only from 9.40 to 9.32 points, so the topic mix does not explain it. Who asks separates outcomes far more sharply than what is asked: acceptance runs from 19.8% for askers below 10 reputation to 66.8% at 10k and above, a 47- point spread with a gradient in every year, against 17 across topics, and a shift toward newer askers accounts for up to 59% of the acceptance fall. The largest of seven properties an asker controls accompanies 5 points. The record places the support gap in the toolchain and among newcomers.

0.38

Superintelligent Software Engineering: Governance That Scales with Capability

cs.SE cs.AI ·2026-10-22 ·Mohammad Mehdi Gholinejad
positioncode/data
Abstract · 16 pages, 6 figures, 6 tables. This version updates citations, corrects a cost figure, marks opinion sentences and clarifies where capabilit…

Software agents are being trained to exceed human engineers; we argue that capability alone does not produce better software engineering. We define superintelligent software engineering as the discipline of building and evolving software with agents whose capability exceeds that of human engineers, in which governance (gates, contracts, decisions and decay control) scales with capability. The argument rests on three claims. First, correctness in a project rests on decisions and contracts that the code does not carry. Second, knowledge decays. Third, once an agent outpaces its reviewers, reading catches a shrinking share of the changes built on stale knowledge. A simple model shows that, past a fixed review capacity, escaping changes grow linearly with throughput while the share that reading catches falls toward zero; governance can lower the slope, capability alone cannot. An exploratory snapshot of 452 of the most-starred public GitHub repositories finds a root-level or .github agent context file in 48.67% (95% confidence interval 44.1 to 53.3); in the median file that names repository paths, 9.5% to 22% of those paths do not resolve, depending on how a reference is read. We then present OSTEON, a discipline for the regime, archived on Zenodo under DOI 10.5281/zenodo.23436020 as a plugin for Claude Code. In OSTEON, each project gets a governed knowledge repository of decisions and contracts, every change is checked mechanically against it, and the knowledge itself is checked for decay at every session start; planning is gated, and sessions record what they learn without writing rules. We report no measured effect of OSTEON, mark claims about its mechanisms as opinion and the author's practice heuristics as design rationale, and state the questions that would falsify the thesis.

0.37

Compact-Bench: Measuring What Context Compaction Costs Long-Horizon Coding Agents

cs.SE cs.AI ·2026-10-22 ·Ruopei Sun, Jianfeng Cai, Kangwen Zhao, Wengang Zhou, Houqiang Li
code/data
Abstract

Context compaction lets long-horizon coding agents continue beyond their context windows, but can erase the state needed to finish a task. Prior evaluations measure efficiency, retained information, or end-to-end success, leaving the effect of a specific history replacement difficult to isolate. Compact-Bench pairs full-history and summary-only continuations from a shared repository state, and evaluates recurring details and successive production compactions. It covers 35 SWE-bench repairs, 146 agentic sessions, and 106 production chains. Across 14 summarizer backbones, stable damage ranges from 5-64% of 22 full-history successes: Claude Opus 4.8 preserves the most repairs, while DeepSeek-Flash damages the most. With Kimi-K3 as summarizer, the Kimi Code prompt preserves the most task continuations (19/22), whereas Claude Code retains the most recurring details (15.0%). LLM summaries retain at most 16.0% of recurring details, compared with 79.1% for mechanical extraction; production summaries retain 15.7% at the first boundary. Failures recur across methods: one repair is damaged by 13/14 backbones, and another by all 5 prompts. These results show that reliable compaction requires preserving the progress needed to complete a task, and that retaining relevant information alone does not ensure successful continuation. Code is available at https://github.com/cai-jianfeng/compact-bench and data is available at https://huggingface.co/datasets/caijanfeng/compact-bench.

0.36

SWE-Clar: Can Coding Agents Clarify Incomplete Client Requirements and Deliver Greenfield Software?

cs.SE cs.AIcs.CL ·2026-10-22 ·Jialei Li, Dingxian Wang, Xiaoyan Zhao, Yimeng Bai, Xuanyi Zhu +9 more
benchmarkcode/data
Abstract · 29 pages, 9 figures, 11 tables. Code and data: https://github.com/upwork/SWE-Clar

Greenfield software development in freelance settings requires coding agents to turn incomplete client requests into working software. This makes requirement clarification essential, yet clients may be unable to provide continuous feedback over long development cycles, and receiving their answers does not guarantee correct understanding of complex requirements. These challenges motivate two benchmark design choices: clarification precedes code implementation, and its effectiveness is assessed through explicit tests of requirement understanding. Guided by these choices, we introduce SWE-Clar, a benchmark grounded in real, high-value client commissions to evaluate requirement clarification and code implementation in this setting. It comprises 100 tasks based on real software development jobs on Upwork, each with at least US\$4,000 in total payments. Each task begins with multi-round interaction between a coding agent and a client agent, followed by separate runs for code implementation with acceptance testing and clarification assessment. For the latter, we construct 512 questions paired with reference answers from real client-freelancer interactions, covering both explicitly stated requirements and those inferred from the conversational context. Both runs begin from the same clarification context but remain isolated, preventing assessment questions from guiding code implementation. Experiments with nine frontier models show that even GPT-6 Astra, which has the highest observed implementation score, achieves a task pass rate of only 67%, underscoring the challenge of independent software delivery in this setting. The dataset and evaluation code are available at https://github.com/upwork/SWE-Clar.

0.35

Tracing Hallucinations in LLM Reasoning for UML Class Diagram Generation: An Empirical Taxonomy and Mitigation Study

cs.SE cs.CL ·2026-10-22 ·Mojdeh Rahmanian, Foutse Khomh, Yanchao Yu, Ashkan Sami
Abstract

Large Language Models (LLMs) can generate UML class diagrams from natural language requirements, but hallucinations may compromise the correctness and completeness of the resulting diagrams. Existing evaluations mostly assess the final diagram, without examining the intermediate traces of reasoning-capable LLMs, which can reveal where these hallucinations arise. We develop an empirical taxonomy of eleven hallucination sub-types organised into three dimensions: Prompt Faithfulness (deviations between requirements and reasoning), Code Faithfulness (discrepancies between reasoning and generated PlantUML code), and Reasoning Quality (problems within the reasoning). The taxonomy was derived from 60 manually annotated DeepSeek-R1 traces, applied to 120 further DeepSeek-R1 traces through a dual-LLM annotation pipeline, and examined on 20 DeepSeek-V4-Pro traces. Code Faithfulness was the most frequent dimension for DeepSeek-R1 in both datasets, whereas Prompt Faithfulness dominated for DeepSeek-V4-Pro, suggesting model-dependent failure patterns. In the manually annotated traces, reasoning-trace length showed no significant monotonic association with hallucination frequency, whereas requirement complexity was negatively associated with it, particularly for Prompt Faithfulness errors. In a mitigation experiment on 13 Golden UML ModelSet cases with three Qwen3-based models, adding Prompt Faithfulness and Reasoning Quality constraints to the generation prompt raised the pooled mean F1 score by 4.0 percentage points, whereas a post-hoc Code Faithfulness review provided no consistent further benefit. Relationship modelling remained the lowest-scoring element throughout. These results suggest that localising hallucinations within the generation process can inform targeted, prompt-based mitigation. Whether these findings generalise beyond the DeepSeek and Qwen3 model families remains to be tested.

0.34

Beyond Correctness: Attribution of Natural-Language Documentation for Formal Libraries

cs.SE cs.AI ·2026-10-22 ·Sirui Jiang, Dongming Xiang, Shichao Wu, Yuxuan Zhou, Yihong Wu +3 more
theory
Abstract · 20 pages, 9 figures, 6 tables

Formal languages are widely used in software engineering to specify and verify program properties. Because their mathematical notation and logic are hard to read, developers often rely on natural-language documentation to understand the functions in formal libraries. Yet a documentation statement can be provably true and still hold after the function's behavior changes. We call a statement's dependence on the function's specific behavior attribution, and we give a method to check it. We design and implement DocAuth, a documentation-generation method that uses attribution to decide what to publish and attaches to each published sentence the formal property recovered from it, together with checkable evidence. On 150 functions from the Community Books library of the ACL2 theorem prover, three large language models produce proved but non-attributable properties for about a third of the functions. DocAuth matches or exceeds the attribution rate of a mutation-based filter while delivering 37% more attributable sentences. Compared with a pipeline that checks truth alone, it raises the share of published sentences that a second model confirms as attributable from 82.9% to 94.7%. These results show that provable truth does not guarantee that documentation reflects what a function actually does. Attribution complements truth as a criterion for generating and evaluating documentation for formal libraries.

0.33

Procrustes-Bering-Sheaft: Automated Model Discovery and Virtual Crash Tests for Microservices

cs.SE cs.DCcs.PF ·2026-10-22 ·Anatoly A. Krasnovsky
Abstract · Submitted to the ICSE 2027 Tool Demonstration and Data Showcase track

Procrustes-Bering-Sheaft automates lightweight model discovery and virtual crash tests for microservices. Each test estimates operation availability by Monte Carlo sampling of service and communication failures. Procrustes checks deployment and tracing evidence; Bering discovers dependencies and incorporates replica declarations; Sheaft evaluates declared operation-success conditions. Independent command-line interfaces exchange reusable, versioned artifacts. The toolchain extends an earlier graph-based approach with mandatory-call contracts, failure-tolerance sweeps, baseline comparison, component-importance analysis, and CI/CD gates. We demonstrate the workflow on one HTTP operation. Six controlled execution states exposed two mismatches between reachability and observed success; an explicit required-call predicate matched all six. Two modelled one-replica additions produced different decisions: recommendation retained an availability-target violation, whereas catalog restored the baseline estimate of 0.979900 and grid tolerance of 0.12. Analytical oracles and controlled tests verify the implementation; archived replay checks numerical continuity with two earlier studies.

0.33

ConQEC: Concolic Testing for Quantum Error-Correcting Code Implementations

cs.SE quant-ph ·2026-10-22 ·Zihao Li, Ji Guan, Mingsheng Ying
top venue
Abstract · 12 pages. Accepted at the 49th IEEE/ACM International Conference on Software Engineering (ICSE 2027)

Quantum error-correcting code (QECC) implementations are hybrid software-control systems: measured syndromes drive decoder dispatch, recovery actions, and logical checks. Existing QECC verification techniques can prove formal obligations, and general quantum testing can expose final failures, but developers still lack a scalable testing workflow that generates concrete faults and explains where a QECC repair path first diverges. This paper presents ConQEC, to our knowledge the first concolic testing framework for QECC implementations. By combining concrete (Pauli) fault replay with symbolic repair-path constraints, ConQEC generates tests that expose not only final logical failures but also fault--syndrome--decoder--recovery (FSDR) diagnostics: the injected fault, observed syndrome, decoder decision, recovery action, oracle result, and earliest divergent stage. Its symbolic encoding of Pauli faults, syndrome predicates, decoder and recovery relations, and oracle checks enables this diagnostic workflow to scale beyond exponentially increasing amplitude-level quantum-state reasoning. We implemented ConQEC and evaluated it on generated source-level QECC families and public QECC artifacts with up to thousands of qubits. ConQEC detects the seeded bug in every experimental case and reports a replay-valid, fault-localizing FSDR witness for each.

0.31

ArcTN: Efficient Tensor-Network Contraction Planning for Quantum Circuit Simulation

cs.MS quant-ph ·2026-10-22 ·Zhengyi Liu, Li Liu, Songtao Liu, Xianglong Liu, Xiangzhen Zhou +5 more
code/data
Abstract · 29 pages, 4 figures, 15 tables. Code and benchmarks: https://github.com/Quill-ArcLight/ArcTN. v2: equal-contribution and corresponding-autho…

Tensor-network simulation of quantum circuits depends on contraction orders with low computational cost and small intermediate tensors, found without excessive planning time. We present ArcTN, a contraction-order optimization framework with two search presets. Light keeps planning overhead low, and Heavy spends more effort on refinement. Both combine randomized-greedy construction, subtree reconfiguration, and parallel tempering with stage-specific stopping rules. ArcTN also provides Fixed and Dynamic slicing to meet tensor-size limits and executes the returned orders. On the 30 QC32 quantum-circuit networks completed by all eight compared configurations, Light reduces FLOPs by 61.32% and plans 1.62$\times$ faster than Cotengra Auto/CMA-ES, and Heavy reduces FLOPs by 43.83% and plans 7.21$\times$ faster than Cotengra AutoHQ/Optuna. Heavy also needs 11.85% fewer FLOPs than OMECO TreeSA-8 and plans 1.33$\times$ faster. Dynamic slicing meets the size limits in all 90 runs on 18 scalar-output networks and reduces total sliced FLOPs by 31.99% relative to Fixed slicing, at the cost of longer planning. With a common NumPy executor, Heavy has the lowest combined planning and execution time among five configurations on the runs that all of them completed. On 672 networks built from 320 MQT Bench circuits, Heavy obtains the lowest geometric-mean FLOPs on the 142 large networks while planning 3.1$\times$ and 31$\times$ faster than Cotengra AutoHQ and Hyper128.

0.31

Mining the Design Space of Multi-Agent Workflows

cs.SE cs.AI ·2026-10-22 ·Xavier Cadet, Junyan Cheng, Peter Chin, Kyle Richardson
Abstract · 21 pages, including appendices. Equal contribution: Xavier Cadet and Junyan Cheng. Website: https://junyan.ch/apex-paper/

Designing large language model (LLM) multi-agent systems (MAS) still relies on trial and error and informal "folk wisdom." Manual studies test only a few communication structures or topologies (e.g., sequential, parallel, shared-memory, or debate), while automated optimizers discard search trajectories after selecting a task-specific workflow. We present a two-phase pipeline for mining reproducible design principles from workflow code and execution traces. We repurpose AFlow's MCTS optimizer across six composable operators and 26 deterministically verified Enigmata tasks, retaining all intermediate rounds to build a corpus of 28,568 executable workflows. A sandboxed AutoResearch loop then mines orchestration code and prompt templates under an $L_0$-penalized validation objective and tests candidate principles on held-out searches. Across 300 feature-mining trials, we identify effects that persist on held-out searches across three levels of workflow design: workflows beginning with independent solvers and one-shot aggregation (Star-first) perform worse ($Δ= -0.045$); persistent shared documents (SharedMemory) outperform debate ($Δ= +0.135$); and prompt specificity and placeholder-based context routing yield gains comparable to communication-structure selection ($Δ= +0.118$). More broadly, our pipeline and results show the feasibility of turning discarded search trajectories into a statistical testbed for discovering and validating multi-agent workflow design principles.

0.30

Typed Template Fuzzing

cs.SE cs.PL ·2026-10-22 ·Lu Maltsis, Dominik Winterer
hot direction
Abstract

Fuzzing software with complex inputs has been a longstanding challenge. Traditionally, fuzzers either programmatically encode language constraints, requiring significant engineering effort, or rely on users to declare grammars and constraints. Recent LLM-based fuzzers address these drawbacks but introduce new issues. Some use LLMs for test generation, sacrificing throughput and producing invalid inputs when the LLM makes mistakes. Others rely on language-specific parsers, static analysis, and AST manipulation libraries. We introduce Typed Template Fuzzing (TTF), a lightweight LLM-based fuzzing methodology. The key idea is to use LLMs in a preprocessing step to transform seeds into typed templates. The typed templates then enable semantically aware, mutation-based fuzzing without language-specific parsers, type checkers, or further LLM queries during fuzzing. By exploiting the correspondence between seeds and templates, TTF can moreover repair LLM mistakes. We have implemented TTF in our tool Typos, which we evaluate across 21 input languages, including C, C++, Go, Rust, SystemVerilog, Scala, and Java, as well as libraries such as PyTorch and Qiskit. In a six-month fuzzing campaign, we reported 135 bugs, 128 of which were previously unknown. Perhaps most notably, on average 96% of Typos's mutants are valid, coming close to the validity rates of domain-specific fuzzers. Our evaluation further shows Typos's benefits over the state of the art.

0.29

From Tokens to Credits to Outcomes: Pricing AI Across Models, Applications, and Cloud Platforms

cs.SE ·2026-10-22 ·Pratik Gupta
Abstract · 26 pages

Between 2023 and 2026, AI changed how software is priced. Traditional SaaS charged per seat because serving one more user cost almost nothing. Model inference has non-zero marginal cost, and usage varies by orders of magnitude across customers, producing prepaid credits, metered usage, hybrid seat-plus-usage models, and outcome pricing that supplement, and in rare cases replace, seat-based pricing. These pricing models have not been systematically characterized. This paper presents a source-traceable study of public pricing artifacts - pricing pages and documentation - across three cohorts: the full Forbes AI 50 (2026) as the AI-native core sample, cloud platforms selling AI services included by a documented rule, and a frozen purposive cohort of 9 distinctive designs. It contributes (1) a dataset coding each offering's pricing primitive and credit mechanics with dated primary sources, prepared for public release with the paper, (2) a taxonomy of the primitives in use and their distribution across cohorts, and (3) evidence-informed design requirements for automated, configurable billing platforms, including ledger state for credits, attribution pipelines for outcome pricing, and configuration support for high pricing-change velocity. Subscription with usage quota is the dominant observable model among AI-native companies (55% of observable AI-native rows; 45-61% under sensitivity analysis of hybrid boundary cases, 41-65% when product-line coding is also varied). 23 of the 52 AI-native rows publish no pricing. Hybrid products are better understood as compositions of packaging anchor, allowance mechanism, and billable unit than as mutually exclusive categories. The dataset is planned for release with the paper; my employer is excluded from the sample.

0.27

The Reversible Intermediate Language L4, its Type System, and its Bidirectional Small-Step Semantics

cs.PL ·2026-10-22 ·Torben Ægidius Mogensen
Abstract · 17 pages

L4 is a generic, reversible, linearly-typed, low-level language intended as an intermediate language for compiling reversible programming languages to reversible machines. L4 is intended to replace and generalise RSSA, a reversible SSA form. Inspiration is drawn from other low-level reversible languages such as HSSA. In this paper, we define L4 first informally, and then give a formal type system and a bidirectional small-step semantics for L4: The rules have direction indicators to determine forwards or backwards execution, and the direction can be changed during execution.

0.26

On U_c-admitting spaces and U_c-completions of T_0 spaces

cs.LO math.GN ·2026-10-22 ·Kangming Lu, Qingguo Li, Shengwen Wang, Rongqi Xiao
theory
Abstract

In this paper, we introduce a new class of spaces, termed U_c-admitting spaces, and investigate their relationships with U_s-admitting spaces and well-filtered spaces. Adopting the perspective of H-sobriety, we show that every U_c-admitting space is H-sober, so that U_c-admitting spaces arise as a special case of H-sober spaces. Consequently, the category U_c-Sob of U_c-admitting spaces and continuous maps is complete and forms a reflective subcategory of Top_0. In addition, we prove that a T_0 space X is U_c-admitting if and only if its Smyth power space is U_c-admitting. Finally, we present the standard U_c-completion of an arbitrary T_0 space.

0.25

Notation3: Graph Terms for RDF

cs.LO cs.AI ·2026-10-22 ·Dörthe Arndt, Pierre-Antoine Champin
Abstract · Submitted to TGDK

Notation3 Logic was introduced as an extension of RDF allowing the user to express rules on top of triples and to make statements about graphs. The original proposal of the logic only came with an informal specification of its semantics which has been refined later on. While this refinement clarified the meaning of rules and implicitly quantified variables, it left the meaning of graphs used as terms - so-called graph terms - open. In this paper we aim to close this gap. We discuss in detail, how graph terms are used in Notation3, which formal requirements they need to meet, and provide examples to illustrate the difficulties when defining their semantics. We then introduce an abstract syntax for the logic and define its semantics in two steps: similarly to RDF's simple interpretations, we define interpretations and basic entailment for Notation3 which covers plain RDF triples but also triples containing graph terms. N3-entailment is then an extension of basic entailment which assigns meaning to the different special predicates of the logic like for example the implication symbol. N3-entailment is defined in a way that it can be extended whenever a new predicate is added to the logic. As Notation3 Logic is currently under development, we expect many predicates to be added. Our research provides and important step towards the full standardisation of the logic.

0.24

Knowing When to Stop: The Evidence-Action Gap in Coding Agents

cs.SE cs.AIcs.CL ·2026-10-22 ·Tianyu Chen, Xinran Zhao, Vijay Viswanathan, Shoubin Yu, Mingyuan Zhou +1 more
position
Abstract · 36 pages

Coding agents are trained to complete tasks, and they keep trying even when the environment makes completion impossible. The evidence that only outside help can restore progress usually appears within a few steps, yet agents rarely act on it: they over-persist, attempt undesired workarounds, and claim success the record does not support -- a phenomenon we call evidence-action gap. To measure it, we introduce ImpasseBench, which injects twelve realistic environment blockers into otherwise solvable repair tasks and marks, in each trajectory, the step at which the collected evidence becomes sufficient. Scoring actions relative to that step separates appropriate effort from over-persistence and premature handoff. Across eight frontier models and 19,371 trajectories, we find that nearly every run that needs outside help collects sufficient evidence, yet a third still exhaust their budget, and every unsafe request we detect comes after that point. Fortunately, an escalation tool, or an external monitor for models that ignore the tool, already guards against much of this behavior: the tool cuts budget exhaustion from 34.0% to 6.7%, and the monitor nearly triples task success after environment restoration. Even then, a fired escalation is not yet a useful handoff: tickets reliably name the blocker but often omit the agent's own progress and verification status. Taken together, we argue that knowing when to stop therefore needs explicit support, both in the tool that lets an agent hand off and in the report it leaves behind.

0.23

MBABench-2.0: Evaluating AI Agents on Constructing Financial Spreadsheets Beyond Simple Verifiability

cs.SE cs.AI ·2026-10-22 ·Julian Poeltl, Thomson Yen, Patrick Shea, Shubham Anil Varma, Dhruv Rastogi +3 more
benchmark
Abstract

Spreadsheets are the lingua franca of modern finance, driving valuation, forecast, audit, and many other core workflows of the industry. As LLM agents advance to automate knowledge work, spreadsheet construction is a natural and pertinent testbed. However, most existing spreadsheet benchmarks evaluate agents on fill-in-the-blank style tasks, where an agent performs minor edits in otherwise complete spreadsheets. This protocol fails to capture the most central and time-consuming part of the spreadsheet workflow: the modeling decisions, format design, and sheet layout regular knowledge workers need to toil through to build a workbook from scratch. We introduce MBABench-2.0, a benchmark of 101 original spreadsheet construction tasks, the hardest of which is estimated to take 33 hours for a modeler with 3+ years of experience. Unlike coding or math, where a unit test or a final answer can verify an agent's output, the quality of a financial spreadsheet that concerns readability, maintainability, or auditability, is not so simply verifiable. To holistically evaluate spreadsheet construction, we introduce 129 criteria spanning 12 categories that operationalize quality beyond the accuracy of the spreadsheet. Evaluating over 20 agents across 8 models and 4 harness types, we find that frontier agents rarely construct spreadsheets that are both accurate and meet our quality criteria: the top agent achieves a pass rate of only 16% on the benchmark. The benchmark demonstrates a substantial gap in spreadsheet building capability of current agents.

0.22

ATLAS: A Multi-Agent Framework for Lifting Binary to Canonical Compiler IR

cs.SE ·2026-10-22 ·Xiaoao Zhu, Ni Zhang, Jie Ren, Zhiqiang Li, Jie Zheng +2 more
Abstract

Lifting stripped, heavily optimized binaries to a canonical, pre-optimization LLVM intermediate representation (IR) is essential for downstream security analysis and legacy software migration when source code is unavailable. However, aggressive compiler optimizations rewrite control and data flow, remove structural metadata, and introduce type-erased low-level idioms, causing current lifters to produce malformed or semantically divergent IR. We present ATLAS, a multi-agent framework designed to reconstruct formally valid and canonical LLVM IR from optimized disassembly.ATLAS combines (i) Structure-aware, region-level retrieval, which utilizes pseudo-probe attribution to align optimized instructions with canonical IR exemplars, and (ii) a collaborative refinement loop that closes the gap between probabilistic generation and deterministic compiler constraints. Concretely, a Lift agent generates an initial IR draft conditioned on retrieved exemplars, an Advisor agent interprets LLVM verifier/compilation diagnostics and static checks to propose localized repair actions, and a Validation agent applies the edits, re-runs toolchain verification and (when available) runtime checks, and feeds the resulting diagnostics back to the Advisor. The agents iterate until the IR is verifier-clean and behaviorally consistent, or a budget limit is reached. On HumanEval-Decompile and MBPP, ATLAS outperforms seven baselines on lifting LLVM IR, achieving 91.8% re-executability and 89.4% CFG similarity across ARM, x86-64, and RISC-V binaries. Moreover, our evaluation demonstrates that ATLAS also outperforms the Binary to C to IR pipelines in pointer-flow precision and downstream analyzability. By recovering a well-formed, machine-centric representation, ATLAS enables practical cross-ISA recompilation and verifiable binary-only migration workflows.

0.22

When Code Gets Cheap, Evaluation Gets Scarce: An Experience Report on AI-Assisted Agile Software Engineering

cs.SE ·2026-10-22 ·Barbara Weber, Olaf Zimmermann, Marco Franceschetti
Abstract

Coding assistants are widely reported to speed up code production; most accounts stop there. We report the end-to-end construction and deployment of a compliance-critical, human-in-the-loop exam-grading system, built by a single developer who wrote no code by hand, protected by a gated process from stakeholder persona to audited behavior. The initial system was built in three working days and could grade a real cohort. We draw one claim from this experience: AI changes software engineering less by making programming faster than by changing what is scarce. Writing code becomes abundant. Reliable evaluation-knowing whether a result is correct and meets the actual need-remains necessary at every iteration and becomes the binding constraint. Engineering judgment remains a second scarcity. Several defects passed the test suite: they sat at the boundaries the tests had mocked, the browser and the language-model API, or appeared only in real use. Deployment exposed assumptions that local tests had missed. Moreover, an AI assistant acting across many steps bypassed process checkpoints until we made them enforceable gates. All of this was observed in a triple union: the developer was also the end user and domain expert, making expert judgment immediately available and evaluation as cheap as it can be. This favors the iteration gains, which may shrink where evaluation requires additional people or expertise, but makes the case a hard test for evaluation scarcity: evaluation still constrained progress. We draw five lessons: run a gated, traceable construction process; specify what is knowable and discover the rest against a running prototype; validate in the real environment; gate the assistant, not only the work; record decisions with their assumptions and the process state the assistant forgets. A second project showed similar patterns; an independent single-expert case corroborates the shift.

0.21

From Natural Language to Verified C: Verification-Guided Generation with ACSL Specifications

cs.SE ·2026-10-22 ·Zian Wang, Qin Li, Weisong Sun, Haibin Cai
Abstract · 12 pages, 1 figure, 7 tables

AI-assisted code generation amplifies the demand for automated formal verification. Generated programs must not only be executable but also accompanied by machine-checkable evidence of correctness. Existing approaches can synthesize functionally plausible code or plausible-looking specifications from natural-language prompts; however, they do not explicitly co-design the implementation and specification as a unified artifact optimized for verification. Consequently, automated verifiers often fail to discharge proof obligations. Typical failure modes include missing or implicit proof anchors, program structures that are hostile to automated reasoning, and inconsistencies between code-level state and specification-level requirement facts arising from the independent generation of code and specifications. Even when verification succeeds, it may only establish weak specifications that omit requirement-critical behavior. This paper introduces VeriGen-C, a formal-verification-friendly framework for the joint synthesis of C programs and ACSL specifications from natural-language requirements. Code generation and repair are constrained to expose proof-relevant state, to streamline control flow for deductive verification, and to eliminate implementation patterns that hinder automated proof search. Specification generation and refinement are similarly constrained to mitigate verification timeouts and to ensure alignment with the user requirements. Empirical evaluations on HumanEval-X, FM-Bench, and LiveCodeBench Lite demonstrate that VeriGen-C substantially improves the rate of verified and requirement-aligned solutions relative to an interface-compatible baseline, from 5.5% to 44.5%, from 44.3% to 80.0%, and from 6.5% to 58.7%, respectively.

0.20

RAIN: Agentic Proof Generation for Rocq-based Software Verification

cs.LO ·2026-10-22 ·Jinyi Zhou, Zenan Li, Yuan Yao, Jun Ma, Taolue Chen +1 more
theory
Abstract

Interactive theorem provers can provide strong correctness guarantees for software, but constructing the proofs remains a challenging and labor-intensive task. Inspired by the recent progress of modern coding agents, we present RAIN, a lightweight proof-specialized agent for Rocq proof generation. RAIN augments an LLM with proof engineering capabilities, including repository exploration, iterative refinement, and checker feedback. This design allows the agent to retrieve lemmas autonomously, revise complete proof scripts instead of isolated tactics, and iteratively refine its proof strategy using enriched proof-state information. Evaluated on a subset from CoqStoq, RAIN shows substantial improvements over the existing baselines, achieving over 90% prove success rate with at least 24.7% improvements. We also compare with general-purpose coding agents, the results show that RAIN maintains comparable prove success rate but with significant lower cost.

0.19

What Open-Source Projects Tell Their Coding Agents: A Census of AGENTS.md and CLAUDE.md

cs.SE cs.AI ·2026-10-22 ·Pawel Zarzycki
Abstract · Corrected abstract metadata formatting and restored missing numerical values; paper unchanged

Coding agents can be guided with instruction files in a repository. We examine this emerging practice through a census of 22,854 popular, active GitHub repositories in thirteen languages, treating each repository's root 'AGENTS$.$md' and 'CLAUDE$.$md' as one instruction set. Agent instructions appear in 26.3% of repositories and 59.6% of the most-starred; where both files exist, 'CLAUDE$.$md' usually points to 'AGENTS$.$md'. We analyse 341,852 statements using model-assisted annotation and human review. The annotations reveal a practical mix of project knowledge and rules about code, building and workflow. Collaboration rules name concrete requirements, from test plans to commit formats, yet 74.0% of directives give no reason. Projects disagree on AI attribution: commit-message rules more often prohibit it than require it, while pull-request rules more often require disclosure. A third of instruction sets have a generator-associated signature, a tool-inserted block, shared text or an AI-marked creation commit; four in five are edited after creation. Prespecified tests associate repository popularity with rules on pull requests and AI contributions. Exploratory analyses link activity to process-oriented content and show that LLM and agent projects are over-represented among adopters. The models agree much more closely with each other than with the human reviewer's topic assignments, a caution for automated annotation. We release the annotations, taxonomies and pointers to the analysed file versions.

0.19

Reuse, Rewrite, or Reject: Evidence-Constrained Trace Reuse for Repeated Mobile GUI Control

cs.SE cs.AIcs.LG ·2026-10-22 ·Bingyu Bai, Cheng Zhang, Erhu Feng
Abstract

Mobile GUI agents frequently perform repetitive tasks with similar interaction procedures. Reusing previous executions can reduce redundant LLM inference and execution overhead, but variations in user instructions and execution environments may invalidate previously successful actions. We systematically characterize these variations into six representative categories and observe that such changes do not necessarily invalidate entire procedures: individual transitions can remain reusable when validated against the current task context and interface state. Based on this insight, we propose graph-text templates for selective execution-trace reuse. The graph captures UI states and action transitions, while the text encodes task intent, procedural structure, and typed slots for variable content. A record-template-replay framework constructs templates from verified executions and leverages the current instruction and live UI state to reuse, adapt, or discard individual transitions. Given an execution trace, our framework reduces LLM inference token consumption by approximately 80% on average across five base agents on AndroidWorld, with most agents achieving at least 1.5-fold execution speedup over LLM-only execution. In real-device evaluations on Android and HarmonyOS, our framework achieves action reuse ratios of 70.7-79.8% while maintaining over 2-fold execution speedup in most evaluated configurations.

0.16

VLTest: Finding Joint Image-Text Failures in Vision-Language Models

cs.SE cs.AI ·2026-10-22 ·Bin Duan, Jintao Lin, Guowei Yang
Abstract

Vision-language models (VLMs) are increasingly embedded in software systems. Testing them raises two problems. Mutating either input can change the expected answer, so an output change is not by itself a failure. Some failures appear only when both inputs are mutated; attributing them to the combination requires testing each mutation alone. We present VLTest, a black-box VLM testing approach. It generates image variants by editing VQGAN latent codes and text variants with a rewrite-expand-prune pipeline, filters variants likely to change the expected answer, and tests each variant alone before testing combinations, so every joint failure has four matched cells. A VLM judge confirms a failure only after checking that the input keeps the expected answer. On three tasks with eight VLMs, including two commercial API models, the oracle confirms 81.7% of the candidates, with 88.8% to 94.6% precision against human labels, and four-cell confirmation rejects 22.5% of joint candidates. Under the same budget, VLTest's generators find 1.8 to 2.5 times as many confirmed failures as SINVAD and TextFooler, and its joint tests find 91 to 197 attributed joint failures per 1,000 joint calls, 1.3 times as many as the same pairing applied to the baselines' variants.

0.15

Folding Pooled Usage Examples into the Tool Index Lowers Retrieval of Unused Tools

cs.SE cs.AI ·2026-10-22 ·Abhijit Chakraborty
Abstract

AI agents often find tools by searching their documentation. Agents that share a tool catalog can also pool usage examples: past requests, each tagged with the tool that handled it. We ask what pooling does to tools no agent has used yet. Adding each usage example to the search index as an extra entry for its tool (folding) helps tools that have examples, but on two of three catalogs it sharply lowers how often a needed unused tool reaches the top five results; on StableToolBench with our main embedding model, the share falls from 73 to 40 percent of queries. When the extra entries hold documentation from other tools instead of requests, the loss is much smaller, though differences in content or writing style could also explain this. We introduce ZoE (Z-scores over Evidence), a scoring rule that scores the documentation and the usage of a tool separately, so a tool without usage keeps its documentation score. On all three catalogs and with both embedding models we tested, ZoE retrieves unused tools more often than a router that reserves result slots for usage matches. With our main embedding model, the router retrieves more of the other tools on two catalogs, and in post hoc comparisons a simpler rule that adds a constant to the scores of tools with usage shows no detected difference from ZoE on unused tools. For five simulated agents on one catalog, folding lowered retrieval of unused tools for every agent; with ZoE no average drop was detected, though some agents still lost. Evaluations of pooled agent experience should report tools no agent has used separately from the average.

0.15

Formal Grammars and Generating Functions: From Algebraic Structure to Non-Algebraic and Non-Derivable Series

cs.FL ·2026-10-22 ·Zhijun Wang
theory
Abstract

This paper investigates how the algebraic and analytic properties of ordinary length generating functions constrain the grammatical and computational complexity of formal languages. We review the Chomsky-Schutzenberger representation, DSV translations for unambiguous context-free grammars, and their stack-indexed extensions, distinguishing finite algebraic systems from more general constructions. Examples and a diagonal argument demonstrate that even a rational generating function need not determine a language's grammatical class. Our central computability result establishes that, for a recursively enumerable language, computability of its length-counting sequence is equivalent to decidability of membership. Consequently, if a recursively enumerable language has a D-finite length generating function over Q(z), it is decidable; equivalently, undecidable recursively enumerable languages cannot have D-finite length generating functions. We then examine coefficient growth and singularities through Pringsheim's theorem, the Polya-Carlson dichotomy, Hadamard gaps, and Siegel G-functions. The resulting picture is asymmetric: analytic properties provide useful one-way constraints on language computability, but length enumeration loses too much information to determine grammatical or computational complexity without additional assumptions.

0.12

CAMGen: A Core Change-Aware Multi-View Framework for Commit Message Generation

cs.SE ·2026-10-22 ·Bo Xiong, Weisheng Chen, Zongen Ren, Yifeng Zhu, Xutian Li +3 more
Abstract · 32 pages, 5 figures, 8 tables, Manuscript submitted to a Journal (2026)

Commit messages help developers understand code changes and support software maintenance. Recent approaches use repository context to improve commit message generation (CMG) with large language models (LLMs). However, existing CMG methods provide limited support for understanding how local edits affect functionality jointly implemented by multiple code entities. This functional understanding is essential for generating commit messages that explain the main change and its impact. To address this problem, we propose CAMGen, a core change-aware multi-view commit message generation framework that incorporates functional context into generation. CAMGen identifies core entities that represent the primary change intent. It then forms candidate functional units through repository-wide semantic clustering and uses those containing the core entities to construct a Functional Semantic View. This view supplies semantic context for interpreting the affected functionality. CAMGen also uses context provided by the Structural Dependency View and the Historical Exemplar View. The context from all three views is then provided to the LLM together with the code diff to guide CMG. The experimental results show that CAMGen outperforms representative CMG baselines across three programming languages. Human evaluation further shows that CAMGen generates more complete commit messages with clearer explanations of change rationale and impact. These results demonstrate the value of functional context and complementary repository information for CMG.

0.10

Theory & Algorithms 51

Deep holes of full-length Reed--Solomon codes: uniform bounds and exact classifications

cs.IT math.COmath.NT ·2026-10-22 ·Xiang Fan
hot direction
Abstract · 43 pages

Let $u$ be a received word for the full-length Reed--Solomon code $\operatorname{RS}_q(k)$, with interpolation polynomial $f_u$ of degree $k+m$. For each $m\in\{3,4,5,6,7\}$, we classify all such deep holes over every finite field, without a field-size hypothesis. None occur for odd $m$ in this range. At $m=4$ and $m=6$, the only exceptions are, respectively, the Segre sextic and Frobenius-power families; these occur only in dimension two over fields of characteristic two. For arbitrary $m\geqslant3$, we also establish two uniform bounds. If $\sqrt q\geqslant12(m+1)\log(2m+2)$, no word of interpolation degree $k+m$ is a deep hole when $q$ is odd, $k\geqslant3$, or $m$ is odd. If $q>16(m+2)^4$, we obtain the corresponding complete classification at every admissible $k$, including the explicit dimension-two exceptions. Our proofs use an agreement criterion expressed in complete homogeneous symmetric functions, agreement-count estimates, finite geometry, permutation-polynomial restrictions, and subset constructions for solving the resulting symmetric-function equations. The remaining finite cases are settled by finite computations with supplied certificates and deterministic verification programs.

0.52

Truthful by Omission: Click Thinning for LLM-Native Advertising

cs.GT cs.AIcs.DS ·2026-10-22 ·Junhua Liu, Zhicheng Liang, Zejun Zhang, Langzhou He, Feng Qian
Abstract

Generative AI assistants are becoming an advertising medium, embedding sponsored content in their answers. This setting departs from classical position auctions in two ways: an ad's click-through rate depends on the passage that hosts it, and because users stop at their first click, earlier ads take clicks from later ones. This competition for clicks makes truthfulness delicate: a higher bid can reshape the allocation and put a competitor ahead of the bidder, reducing the bidder's clicks. Existing mechanisms either attain a constant fraction of the optimal welfare (advertisers' total click value) but are not known to be truthful, or are truthful but their welfare guarantee degrades as the number of passages grows. We achieve both even when the answer and its editorial order are fixed before bidding. Assuming each ad's click-through rate when reached is known and unaffected by omitting other ads, we give a computationally efficient mechanism that is truthful in expectation and attains at least half the optimal welfare under free reordering. Our key idea, click thinning, omits ads at random, showing earlier ads less often so that each ad's overall click probability no longer depends on the order of the ads. Welfare maximization then reduces to a tractable matching problem, and solving it exactly with VCG-style payments makes truthful bidding optimal in expectation. Characterizing the click probabilities that omission can achieve sharpens the approximation factor from $2$ to $2-(1-δ)^{K-1}$ when at most $K$ ads are shown and each click-through rate is at most $δ$. This factor is the best possible for any order-preserving rule, even one that ignores incentives and computation, so truthfulness and computational efficiency incur no additional worst-case loss.

0.48

No Perfect Randomized Composition

cs.CC math.CO ·2026-10-22 ·Alexander Meiburg
Abstract · 56 pages

Does computing a composition of total Boolean functions require the product of the randomized query complexities of its factors? We give a negative answer. There are families of total functions $f,g$ such that $R(f\circ g) =Θ\!\left(\frac1{\log R(g)}\right)R(f)R(g)$ This result requires first separating linearized query complexity $\LR$, as well as noisy randomized query complexity $noisyR$, from standard randomized query complexity $R$. Both constructions essentially come from totalizing existing partial function separations, albeit via different techniques. By adapting these techniques, we also construct total functions separating every adjacent pair in the chain from sabotage $RS$, max-conflict $\barχ$, and linearised query complexity $LR$. We show that AND-of-ORs separates conflict $χ$ from max-conflict $\barχ$ complexity and makes conflict and sabotage incomparable. Finally, randomized complexity can be smaller than its composition limit $R^*$ by the maximal factor $Θ(\log R)$. The total--total composition separation is also formalized in Lean.

0.48

Extension of MDS Codes over Finite Fields

cs.IT ·2026-10-22 ·Puyin Wang, Jinquan Luo
hot direction
Abstract

In this paper, we investigate extending of $[n,k,n-k+1]$ non-Reed--Solomon(non-RS) MDS codes to $[n+1,k,n-k+2]$ non-RS MDS codes. We develop a practical iterative algorithm for constructing longer MDS codes while preserving their non-RS structure. The proposed algorithm is inspired by the meet-in-the-middle algorithm in cryptanalysis. On average, it requires at most $O(kq^k)$ multiplications, compared with $O\left(\binom{n}{k-1} q^{k-1} k^3\right)$ multiplications for generic exhaustive search. Applying the proposed algorithm to several known families of MDS codes, including elliptic-curve, twisted Reed--Solomon, column-twisted Reed--Solomon, cyclic, and missing-monomial MDS codes over fields of order at most $32$, we obtain new MDS codes of increased lengths that remain inequivalent to Reed--Solomon codes. In particular, we construct $[9,4,6]$ codes over $\F_9$ and $\F_{13}$, each with Schur-square dimension $9$. Over $\F_{17}$, we obtain two inequivalent $[10,4,7]$ codes with Schur-square dimensions $9$ and $10$, respectively. The latter is the dual of an extended $[10,6,5]$ code, demonstrating that the method also applies when $k>n/2$. For these dimension-four codes, the Schur-square dimensions $9$ and $10$ exceed both that of RS code (Schur-square dimension $7$) and those of specific elliptic-curve and column-twisted Reed--Solomon families (Schur-square dimension $8$). Some of these parameters of non-RS MDS codes have not been reported before.

0.47

Report the tolerance, not the count: exact phylogenetic distance deconvolution with incomplete references

cs.DS q-bio.PE ·2026-10-22 ·Haotian Tang
code/data
Abstract · 13 pages, 3 figures. Supplementary appendices (37 pages) are included as an ancillary file. Code, exact input fixtures and saved run records…

Phylogenetic distance deconvolution represents mixture-to-reference distances by weighted sources placed on a reference tree. We study the minimum number of strictly interior sources that explains a distance profile, and what incomplete references cost. On a positive-rational reduced tree we give an exact algorithm with 2^O(K) poly(B) bit complexity for a free positive mean pendant and any nonempty set of observed leaves; the single-exponential dependence on K is tight under ETH, already with every leaf observed. With arbitrary missing references and a prescribed pendant the problem is W[1]-hard in the source budget K, even on positive binary trees supplied with a strict explanation using at most K+1 sources. We then use the algorithm as an instrument. Removing references can only lower the minimum source count, never raise it, at every error tolerance. Which references are missing then matters far more than how many. Under nearest-to-source deletion a median of two references can be erased before the generating support stops being the unique smallest explanation, and that budget stays nearly flat as the panel grows (Spearman 0.10 over 20 to 361 references, against 0.82 for random loss). Discarding 7,931 of 10,575 references from a real 16S panel at random costs 1.8% of the tolerance at which one source suffices. Discarding comparably many as whole clades costs 90%. That tolerance needs neither ground truth nor an error model, which makes the model measurable on real data. The generating configuration's worst-leaf residual exceeds the one-source optimum in all 296 real ANI cases, and in 248 after refitting the common pendant. The phenomenon is not unobservable in practice. A point estimate of the count alone does not quantify it. Methods for this problem should report the tolerance at which an explanation of a given size exists.

0.43

New lower bounds for kissing numbers in dimensions 32-35 and 37-39

cs.DM math.MGcs.ITmath.CO ·2026-10-22 ·Qihong Lin
code/data
Abstract · 8 pages. Certificates and verification programs: https://github.com/QH-Lin/kissing-number-32 (doi:10.5281/zenodo.23527725) and https://githu…

We show that the kissing numbers in dimensions 32, 33, 34, 35, 37, 38 and 39 are at least 352320, 370240, 390980, 415308, 499816, 605322 and 768916, improving the previous bounds 347584, 364736, 384836, 409676, 498024, 591900 and 763668. Except in dimension 38, the configurations are those of Edel, Rains and Sloane with new binary constant-weight codes of weight 8 and minimum distance 8 in the middle layer; each additional word adds 128 points. In dimension 38 we add 6855 antipodal pairs of vectors of the second shell of the Leech lattice to the empty equator of Kravatskiy's 591612-point configuration, where 144 pairs were known; this is an independent set problem in a regular graph of degree 11178 on 8386560 vertices. The codes and lines were found by local search, and all configurations were checked pair by pair in exact integer arithmetic.

0.40

A Sharp Analysis of Fixed Offset Hash Tables

cs.DS ·2026-10-22 ·Yuhao Guo, Seth Pettie, Chengzhang Wan
Abstract

We consider the class of open addressing hash tables that uses a fixed offset sequence $r_0,r_1,\ldots,r_{n-1}$ to find an open slot in the table. As special cases this subsumes linear probing ($r_i=i$) and quadratic probing ($r_i=i^2$). We give a simple analysis that the expected insertion time when the load factor is $1-ε$ is $O(ε^{-2})$. This bound is tight for linear probing (but not for quadratic probing), so the analysis cannot be improved at this level of generality.

0.40

Two-Timescale Antenna Positioning for Pinching-Antenna System Based on Statistical Channel Information

cs.IT ·2026-10-22 ·Jingze Ding, Zijian Zhou, Bingli Jiao, Rui Zhang
top venuetheory
Abstract · This paper has been accepted by IEEE Transactions on Communications

This paper introduces a two-timescale antenna position optimization framework for the pinching-antenna system (PASS). The proposed framework exploits statistical channel state information (CSI) to determine large-timescale pinching-antenna (PA) positions over the entire waveguide, and utilizes instantaneous CSI to refine PA positions at the small timescale within a local region. This approach effectively captures the benefits of large-timescale PA positioning for long-term system performance and small-timescale PA movement for real-time adaptation to fast channel fluctuations. For the single-user case, we derive closed-form solutions for the optimal large-timescale PA positions, and further analyze a performance upper bound together with the conditions for achieving this bound, which guide the small-timescale position refinement. For the multi-user case, an upper bound on the minimum rate among all users is derived. To maximize their minimum rate, we develop an element-wise search algorithm for large-timescale PA positioning and a projected gradient ascent method for small-timescale position refinement. Simulation results show that the proposed scheme closely approaches the derived bounds and consistently outperforms conventional schemes by reducing path loss via large-timescale PA positioning and suppressing inter-user interference through small-timescale local movement.

0.39

Two-colour crossing minimization in complete geometric graphs is NP-hard

cs.CC cs.CG ·2026-10-22 ·Zhe Zhang, Yi Liu, Dan Yang
theorycode/data
Abstract · 34 pages, 5 figures, 2 tables. Updated author information, presentation and declarations. Proof code and reproducible computations: https://…

Given a fixed straight-line drawing of a complete graph, we seek a two-colouring of its edges that minimizes monochromatic crossing pairs. We prove that the threshold decision problem is NP-complete, even for points in general position with integer coordinates. Thus exact minimization is NP-hard, resolving the complexity question posed at the Crossing Number Workshop in 2019. We reduce from MaxCut on permutation graphs by representing source vertices with classes of geometric edges, whose crossings encode source adjacencies. The target complete graph also contains every other edge between its points, and their freely chosen colours can affect the encoded cost. To control these interactions, we construct a large comparison grid with two stable colour phases and use program points to turn this reference into colour preferences on other edges. The optimum differs from a computable baseline plus the amplified minimum number of uncut source edges by an error small enough to distinguish consecutive source values. The reduction is deterministic, with polynomially many points and polynomial coordinate bit length.

0.38

Differentiable Combinatorial Auction Design: Optimizing Revenue-Welfare-Fairness Frontiers With Graph Attention Networks

cs.GT ·2026-10-22 ·Mai Pham, Vikrant Vaze, Peter Chin
Abstract

Designing optimal auctions for allocating public infrastructure-such as airport slots and energy resources-is challenging due to interdependent resources and complex side-constraints. While the Vickrey-Clarke-Groves (VCG) auction guarantees maximum welfare, it cannot maximize revenue or accommodate the multi-objective trade-offs inherent in public infrastructure systems. We propose a differentiable, constraint-aware framework for combinatorial auction design that enables end-to-end learning of fractionally feasible, approximately strategy-proof mechanisms. Our primary contribution is a combinatorial constraint enforcement technique that ensures fractional feasibility by construction, bypassing the need for penalty-based approximations. We introduce CAGraph, a novel graph-based architecture designed to capture dependencies between items and bidders, alongside CANet and CAFormer as deep learning baselines, for combinatorial settings. We also provide a Matching Pursuit-inspired decomposition that maps fractional allocations to feasible discrete allocations and empirically evaluates the resulting discretization gap. In synthetic benchmarks and an airport slot-divestiture case study, respectively, the learned mechanisms achieve up to 73.5% and 180% higher expected revenue than VCG, under small empirical regret. Finally, we visualize the Pareto frontiers for revenue, welfare, and fairness, providing regulators with a decision-support tool to design infrastructure auctions that balance profitability, efficiency and equity.

0.38

Can Reciprocal RIS Perform as Non-Reciprocal RIS?

cs.IT eess.SP ·2026-10-22 ·Matteo Nerini, Juan Vidal Alegría, Bruno Clerckx
publishedunusual
Abstract · To be presented at Asilomar 2026

Yes, if and only if some optimal non-reciprocal RIS (i.e., maximizing a given performance metric) makes the round-trip channel transmitter-RIS-transmitter symmetric, which holds if the channel transmitter-RIS has rank one or if single-stream transmission is optimal.

0.37

Near-Quadratic Time Multivariate Polynomial Interpolation

cs.DS cs.CCcs.SC ·2026-10-22 ·Chandra Kanta Mohapatra
Abstract

Let $\F$ be a finite field. Given $N$ distinct points $\bmα_1,\ldots,\bmα_N \in \F^m$ and associated values $β_1,\ldots,β_N \in \F$, we study the problem of computing an $m$-variate polynomial $f(\vy) \in \F[\vy]$ of individual degree less than $d$, with $d^m \ge N$, that satisfies $f(\bmα_i)=β_i$ for all $i\in[N]$, whenever such an $f$ exists. This is known as the\textit{ multivariate polynomial interpolation} problem. In this work, given $N=d^m$, we present a deterministic algorithm for multivariate polynomial interpolation over an arbitrary finite field $\F$ that runs in near-quadratic time $N^{2+o(1)}\poly(m,d,\log|\F|)$, assuming $d$ is $ω(1)$. This bound holds whenever the associated \emph{multivariate Vandermonde} matrix is nonsingular, has constant co-rank, or admits some special structure. In the general case, we give a randomized Las Vegas algorithm attaining the same near-quadratic running time, provided the finite field is sufficiently large and $d$ grows similar to the previous case. Unlike other multivariate interpolation strategies, we do not assume black-box access to the polynomial. In particular, we are not permitted to query arbitrary input points to obtain their evaluations. Our interpolation result follows by combining Wiedemann's classical approach~\cite{w86} for solving sparse linear systems with known bounds for fast \emph{multivariate multipoint evaluation} over finite fields~\cite{umans08, KU11, BGKM23, BGGKU24}. Specifically, our deterministic interpolation strategy relies on a deterministic variant of Wiedemann's approach, which is more commonly used in the randomized setting.

0.34

Pigeonhole Equal Subset Sum: Subexponential Algorithm, Tight Lower Bounds, and Average-Case Analysis

cs.DS ·2026-10-22 ·Deepak Bhati, Pranjal Dutta, Antoine Joux, Mahesh Sreekumar Rajasree, Karol Wegrzycki
top venuetheory
Abstract · 66 pages, 2 figures, published in FOCS 2026

The Equal Subset Sum problem asks, given n positive integers bounded by M, to find two distinct non-empty subsets with the same sum. We study this problem in the regime $M = 2^{o(n)}$, where the pigeonhole principle guarantees that a solution exists, and the main challenge is to find one efficiently. A straightforward dynamic-programming algorithm runs in time poly(n).$M$. After a sequence of improvements, the current best algorithm, due to Jin, Williams, and Zhang (ESA 2025), achieves running time poly(n).$M^{1/3}$. In this work, we give the first poly(n).$M^{o(1)}$-time algorithm for Equal Subset Sum in the regime $M = 2^{o(n)}$. We complement this algorithmic result with a tight conditional lower bound showing that any substantial improvement would require a major breakthrough in approximation algorithms for lattice problems, building on recent techniques of Vafa and Vaikuntanathan (STOC 2025). Finally, we show that our techniques yield an asymptotically faster algorithm for the regime when $M = Θ(2^{n})$: a poly(n)$\cdot 2^{n/4}$-time algorithm in the average case.

0.32

Matching Tasks to Agents: Mechanism Design for Multi-Agent Contracts

cs.GT ·2026-10-22 ·Matteo Bollini, Matteo Castiglioni, Alberto Marchesi
benchmark
Abstract

We study task assignment under adverse selection and moral hazard. In the single-task setting, agents privately know their types, which determine their success probabilities and effort costs. Based on the agents' reported types, the principal selects an agent to perform the task and offers a success-contingent payment. Although optimal Bayesian contract design is generally computationally intractable and Myerson's characterization does not directly extend to settings with hidden actions and multidimensional private information, we show that the single-task problem admits a simple Myerson-style characterization. Our key insight is that incentive constraints depend only on the ratio between effort cost and success probability. We characterize an optimal mechanism and show that it can be computed in polynomial time. We also establish a tight $Θ(1/(NK))$ worst-case bound on the ratio between the optimal profit and the full-information benchmark, where $N$ is the number of agents and $K$ is the number of types per agent. We then extend our analysis to multiple tasks, showing that an optimal mechanism may not exist. Nevertheless, for a constant number of agents, we provide a polynomial-time approximation scheme for the supremum profit.

0.32

Temporal Degeneracy Limits Message Passing Decoding of QLDPC Codes Under Repeated Noisy Syndrome Measurements

cs.IT quant-ph ·2026-10-22 ·Dimitris Chytas, Nithin Raveendran, Bane Vasić
Abstract

Practical quantum computations can require logical error rates of $10^{-12}$ or lower. Quantum low-density parity-check (QLDPC) codes are promising candidates for reaching such rates with low qubit overhead, provided that their syndromes can be decoded efficiently. Repeated noisy syndrome measurements introduce temporal degeneracy into the space-time decoding graph. We characterize the temporal kernel, whose vectors have zero total data parity, and show that it is generated by elementary temporal codewords of weight $d_v+2$, where $d_v$ is the variable-node degree. For $(d_v,d_c)$-regular parity-check matrices with fixed degrees $d_v,d_c>2$ and girth greater than four, we derive counting bounds and construct degenerate error patterns of weight $(d_v+2)/2$ for even $d_v$ and $d_v+1$ for odd $d_v$, with multiplicity $Θ(nr)$ in the blocklength $n$ and number of measurement rounds $r$. Under the isolation assumption, these patterns cause local failures of deterministic isotropic message-passing decoders under a flooding schedule. These failures depend on the node degrees rather than the minimum distance; for the bivariate bicycle (BB) codes considered, their weight is equal to four. Monte carlo simulations and fixed-weight importance sampling quantify harmful low-weight degenerate error patterns for the min-sum (MS) decoder beyond the isolation assumption. Motivated by the identified symmetries, we propose space-time anisotropic min-sum decoding with syndrome-guided realizations. Its complexity is linear in the number of Tanner graph edges for fixed maximum number of iterations and number of decoders. Numerical results on BB codes show threshold-like behavior acrossthe tested code sizes and improved failure probabilities relative to MS decoding with ordered statistics decoding post-processing.

0.31

An integer determinant approach to the length of Type II codes and the weights of shadows

cs.IT ·2026-10-22 ·Jae-Hyun Baek, Jon-Lark Kim
theory
Abstract · 10 pages

It is well known that the length $n$ of a binary Type II code is divisible by $8$. The classical proof relies on Gleason's theorem on weight enumerators. Motivated by a question of Vera Pless, we give elementary proofs of this fact that avoid invariant theory. Our main tool is a congruence modulo $8$ for the determinant of the integer Gram matrix $AA^{\mathsf T}$ of a binary orthogonal matrix $A$, obtained directly from the Leibniz expansion. It gives a short proof of a theorem of Janusz: the number of rows of weight congruent to $3 \pmod 4$ in a binary orthogonal matrix is divisible by $4$. From this single statement we derive both the length restriction for Type II codes and the theorem of Conway and Sloane that every vector in the shadow of a binary self-dual code of length $n$ has weight congruent to $n/2 \pmod 4$. We also give two short character-theoretic proofs of the length restriction and explain how all of these arguments detect the Arf invariant of a quadratic form.

0.30

Temporal Liquidity Crossing for Blockchains: An EIP-1559 Extension for Admission Below the Base Fee

cs.GT cs.CRcs.DC ·2026-10-22 ·Duanyang Dan Guo
Abstract · 28 pages

Transactions differ in how their value changes with execution time and relative position. Ethereum's EIP-1559 lets them compete through fee fields, but a fee bid gives no enforceable claim to an earlier position in the block, and no way to express willingness to accept a later one. Its base fee is a hard admission floor. We introduce temporal liquidity markets (TLM), a framework for crossing complementary temporal requirements: transactions seeking early-region execution (consumers) can fund transactions that opt into exclusion from that region (providers). Selected consumers complete within the early region and are ordered among themselves by authorization density, both validator-checked. In our block-local EIP-1559 extension, providers may be admitted below the base-fee floor. Under TLM regulation, funding-conditioned and capacity-capped provider gas is excluded from the ordinary EIP-1559 controller input. Counting it there instead raises the base fee and displaces ordinary demand at a positive base-fee fixed point. Exclusion lets the controller regulate ordinary demand at its original target and preserves the no-TLM ordinary allocation by construction; in synthetic experiments it raises total utilization with funded gas. Selected consumers pay their signed authorizations, which fund provider shortfalls; the surplus is burned. We state an adversary model with strategic senders and builders and prove eight properties under it, among them validator-checked rule validity, full burn of the base fee, and no subsidy. An inserted consumer bid cannot raise any other consumer's charge. With a provider fee-cap floor, a small, capped inclusion fee and a consumer allowance tied to funded provider gas by a large enough link ratio, no finite set of consumer bids or providers that the builder inserts raises its direct fee payoff.

0.29

Energy-Efficient UAV Covert Communications via Metaheuristic Optimization

cs.IT ·2026-10-22 ·Nguyen Minh Quan, Tran Thanh Thu, Nguyen The Ngoc Khue, Ninh Thi Thanh Tam, Trinh Van Chien
publishedbenchmark
Abstract · 6 pages, 6 figures. Presented at ATC 2026

Wireless communication systems increasingly require not only confidentiality but also transmission undetectability, motivating covert communications. This paper studies an unmanned aerial vehicle (UAV) enabled covert communication system where a UAV serves a legitimate receiver in the presence of multiple active and colluding wardens with imperfect location information. We maximize covert energy efficiency (CEE) by jointly optimizing the UAV's 3D trajectory and transmit power while accounting for energy. To address the resulting non-convex problem with robust covertness constraints, we develop Genetic Algorithm (GA) and Particle Swarm Optimization (PSO)-based metaheuristic algorithms and compare them with a Block Coordinate Descent with Successive Convex Approximation (BCD-SCA) benchmark. Numerical results show that the proposed methods improve CEE while satisfying covertness constraints.

0.29

VisFlow: Interpretable Flow Matching for Visibility Reconstruction, Characterization, and Recognition

cs.CG cs.CV ·2026-10-22 ·Rahul Moorthy, Volkan Isler
Abstract

Methods for learning continuous representations, such as latent manifolds, are fundamental and ubiquitous in modern machine learning. Yet many geometric objects combine continuous shape information with discrete combinatorial structure. This combination presents a fundamental challenge for representation learning because small geometric perturbations can induce abrupt changes in the underlying discrete representation. As a result, the smoothness assumptions that make continuous representations effective no longer hold. Polygons provide a canonical example. Their vertex locations vary continuously, but infinitesimal changes in geometry can dramatically alter the associated visibility graph. Existing methods address the inverse mapping from visibility graphs to polygon geometry using vertex-, triangulation-, or signed-distance-function (SDF)-based representations. Vertex- and triangulation-based approaches construct polygons through explicit geometric operations. However, they are less effective at modeling the relationship between geometry and visibility. SDF-based approaches capture this relationship more naturally. Their intermediate states, however, do not correspond to explicit stages of polygon construction. We present VisFlow, a novel, interpretable flow formulation that combines continuous SDF representations with explicit polygon evolution. We evaluate VisFlow on Visibility Reconstruction and Visibility Characterization. Visibility Reconstruction generates one polygon consistent with a visibility graph (G), while Visibility Characterization characterizes all such polygons. VisFlow maintains 20% more visibility relations on Visibility Reconstruction and improves coverage by 30% on Visibility Characterization compared to state-of-the-art methods.

0.28

Bridging persistence of discrete and continuous 1D maps: Total variation, permutation equivalence, and diagonal affine transformations

cs.CG math.AT ·2026-10-22 ·Matthew Blake, Firas A. Khasawneh
Abstract · 60 pages, 24 figures

Fundamental to the field of topological signal processing (TSP) is the \textit{persistence diagram} (PD), or barcode, of a signal, which gives noise-resistant measures of topologically significant features as distinct from noise. Existing literature proves that the total persistence computed from these diagrams is equal to the total variation, for certain continuous signals on compact domains. We provide a closed formula for this shared quantity, and generalize to all continuous and discrete functions on bounded domains, and to certain functions on non-compact 1D manifolds. Extending the identification of continuous and discrete persistence, we investigate the similarities of functions with the same PDs, and prove that, for any $n$, the persistence diagram can be calculated as a function of the sequence of critical values; we compute these formulas for $n \leq 6$. Finally, we examine which function transformations cause predictable changes in the PD of the transformed function, characterizing affine and linear transformations.

0.28

Sum-rank analogues of Reed--Muller codes over Abelian Galois extensions

cs.IT math.COmath.RA ·2026-10-22 ·Alessandro Neri, Giuseppe Zampella
Abstract

Let $\mathbb L/\mathbb K$ be a finite Abelian Galois extension with a fixed decomposition $\mathrm{Gal}(\mathbb L/\mathbb K)=\langleσ_1\rangle\times\cdots\times\langleσ_m\rangle$. We show that the quotient of the multivariate skew polynomial ring $\mathbb L[x_1,\ldots,x_m;σ_1,\ldots,σ_m]$ by suitable central polynomials $H_1(x_1),\ldots,H_m(x_m)$ is isomorphic to a product of $t$ copies of $\mathrm{End}_{\mathbb K}(\mathbb L)$, so that one skew polynomial represents all blocks of a sum-rank metric space. The sum-rank weight of an element equals the rank of an associated Dickson matrix and of a generalized Moore matrix. Using this, we define sum-rank Reed--Muller-type codes and show that their dimension and minimum distance are given by the formulas for affine Cartesian codes with side lengths $\mathrm{deg}\,H_1,\ldots,\mathrm{deg}\,H_m$; the same holds for codes supported on lower sets. The family contains as special cases linearized Reed--Solomon codes, Gabidulin codes, rank-metric Reed--Muller codes, and affine Cartesian codes. Over finite fields, every linearized Reed--Muller code of Berardini and Caruso is, after a monomial change of variables, a monomial code in our framework. This gives the dimension of all these codes, and their exact minimum distance in every degree whenever the support is a lower set.

0.28

The Cheng-Murray Conjecture on Deep Holes of Full-Length Reed-Solomon Codes

cs.IT ·2026-10-22 ·Weijun Fang
theoryhot direction
Abstract · 33pages

Let $RS_q(k)$ denote the full-length Reed--Solomon code over $\mathbb{F}_q$. Every word $u=(u_x)_{x\in \mathbb{F}_q} \in \mathbb{F}_q^q$ has a unique interpolation polynomial $U(X)\in \mathbb{F}_q[X]$ of degree less than $q$, satisfying $U(x)=u_x$ for all $x\in \mathbb{F}_q$. It is well known that every word with $°U=k$ is a deep hole. The Cheng--Murray conjecture asserts that these are the only deep holes for $3\le k\le q-2$, except when $q$ is even and $k=q-3$. Previous results established the conjecture for $k\le p-1$ and for $k\ge\lfloor(q-1)/2\rfloor$, where $p$ is the characteristic of $\mathbb{F}_q$. In this paper, we prove the Cheng--Murray conjecture in the remaining range $3\le k\le\lfloor q/2\rfloor$, thereby resolving it completely. Our proof combines two complementary approaches. The tangent envelope of a marked arc yields differential constraints that establish the conjecture for $3\le k\le q/2$ when $q$ is even, and for $3\le k\le\lceil q/3\rceil$ when $q$ is odd. For the remaining cases with $q$ odd, we combine local syndrome compatibility with combinatorial arguments to rule out additional deep holes.

0.28

A new lower bound for the sixth Schur number: $S(6)\ge 537$

cs.DM math.CO ·2026-10-22 ·Zhiyang Chen, Ligeng Zhu, Fan Long
theory
Abstract

We give an explicit partition of $\{1,\ldots,537\}$ into six sum-free sets, establishing $S(6)\ge 537$. This improves the lower bound 536 of Fredricksen and Sweet (2000). Known recursive constructions then give the improved lower bounds $S(8)\ge 5\,372$, $S(11)\ge 204\,208$, $S(13)\ge 2\,042\,082$, and $R_6(3)\ge 539$. We also report 131 pairwise inequivalent partitions found by the search. We describe their structure and provide the complete partitions in the supplementary material. These partitions provide starting points for future searches and structural studies of Schur partitions.

0.27

Homomorphic Encryption with Coset Code Construction

cs.IT cs.CR ·2026-10-22 ·Haodong Yang, Venkata Gandikota
theory
Abstract · 28 pages

Homomorphic encryption enables computation on input data without decrypting it. This paper extends a construction based on scalar message codes, utilizing coset codes to pack $k$ message symbols into a ciphertext vector of length $n$. For a given field and vector length, this packing method reduces the number of ciphertext vectors required to transmit a message record. A compatible family of evaluation codes supports public addition and multiplication operations up to a predetermined degree. We prove that the scheme achieves bounded chosen-plaintext security under the coordinate-label distinguishability assumption.

0.27

A Resource Allocation Tournament

cs.GT math.OCecon.TH ·2026-10-22 ·Adam Nathaniel, Refael Hassin
Abstract

We introduce and analyze a round-robin simultaneous resource allocation contest, motivated by sports leagues and other competitive environments where agents must distribute limited resources across multiple encounters. Players allocate resources in advance to matches against all opponents, and match outcomes are determined by relative allocations. We study two payoff schemes: total points, where players accumulate rewards for each win, and winner-takes-all, where only the tournament winner is rewarded. For three players, we provide complete equilibrium characterizations under both schemes. For four or more players, we show that the total points payoff admits Blotto-type equilibria, while the winner-takes-all payoff requires numerical and algorithmic analysis. To address this, we develop a best-response dynamics algorithm with stochastic inertia and use it to approximate equilibria in games with four and five players.

0.26

Kempe Reachability with $k\geq 3$ Colors Is $\mathsf{PSPACE}$-Complete on Claw-Free Graphs

cs.DS cs.DM ·2026-10-22 ·Duc A. Hoang
Abstract

A \emph{proper vertex coloring} assigns colors to vertices so that adjacent vertices have distinct colors. A \emph{Kempe change} in a proper vertex coloring exchanges two colors on one component of the subgraph induced by the vertices of those colors. For a fixed palette of $k$ colors, \textsc{Kempe Reachability} asks whether two given proper vertex colorings of the same graph are connected by a sequence of Kempe changes. This problem is known to be $\mathsf{PSPACE}$-complete for every fixed $k\geq 3$. A graph is \emph{claw-free} if no vertex has three pairwise nonadjacent neighbors. The \emph{line graph} of a graph has one vertex for each edge, with two vertices adjacent exactly when the corresponding edges share an end. Line graphs are claw-free, and their Kempe changes correspond to \emph{edge-Kempe changes} in the original graph, which exchange two colors on one component of the subgraph formed by the edges of those colors. A claw-free graph with a proper $k$-coloring has maximum degree at most $2k-2$, whereas for $k\geq 4$ all proper $k$-colorings of a graph of maximum degree at most $k$ are connected by Kempe changes. We show that, for every fixed $k\geq 3$, \textsc{Kempe Reachability} with $k$ colors is $\mathsf{PSPACE}$-complete even on line graphs of connected simple $k$-regular graphs, and hence even on connected $(2k-2)$-regular claw-free graphs. Equivalently, deciding whether two proper $k$-edge-colorings are connected by edge-Kempe changes is $\mathsf{PSPACE}$-complete even on connected simple $k$-regular graphs. ChatGPT 6 Pro and Claude Fable 5.1 were used to assist with developing and reviewing mathematical arguments and drafting and revising the paper. The author takes full responsibility for the correctness and content of the paper.

0.26

Chordal Construction of Column-Twisted Reed--Solomon Codes

cs.IT ·2026-10-22 ·Anuj Kumar Bhagat, Krishna Kaipa
hot direction
Abstract

A $[n,k]_q$ Generalized Reed--Solomon (GRS) code can be represented, up to monomial equivalence, by a projective system consisting of $n$ points on a degree $(k-1)$ rational normal curve (RNC) $C_{k-1}$ in PG$(k-1,q)$. We use the term \textit{Column-Twisted Reed--Solomon (CTRS)} codes broadly for $[n,k]_q$ codes whose projective system contains at least $(k+2)$ points on a $C_{k-1}$ of PG$(k-1,q)$. By a chord of $C_{k-1}$, we mean a line which is either a tangent, a real secant, or an imaginary secant of $C_{k-1}$. Roth and Lempel constructed non-GRS MDS $[n,k]_q$ codes by considering projective systems consisting of $(n-1)$ points on $C_{k-1}$ together with a point on a tangent line. More recently, Liu et al. obtained non-GRS $[n, k]_q$ MDS codes by considering projective systems consisting of $(n-2)$ points on $C_{k-1}$ together with a pair of points on a real secant. In the case when $q$ is odd, the length of these codes satisfies $k+4 \leq n \leq (q+3)/2$. In this article, we construct non-GRS $[n,k]_q$ MDS codes by considering projective systems consisting of $(n-2)$ points on $C_{k-1}$ together with a pair of Galois conjugate points on an imaginary secant. For odd $q$, these codes have lengths satisfying $k+4 \le n\le (q+5)/2$, thereby improving both the attainable lengths and dimension ranges of previous TRS and CTRS constructions. In particular, our code attains the length $(q+5)/2$ for every $3\le k\le (q-3)/2$. We use a geometric argument to establish the non-GRS property of CTRS codes for the range $3\le k \le n-4$, while the traditional method of comparison with the Schur square of GRS codes proves the non-GRS property only for the narrower range $3\le k \le n/2$. We also determine the structure of the dual of any CTRS code.

0.25

Beyond Homology: Non-Abelian and Orientability Invariants Sensitivity in Cellular Networks

cs.CG ·2026-10-22 ·Raghuwansh Raj, George Panagopoulos, Jun Pang
benchmark
Abstract

Topological deep learning (TDL) architectures are typically evaluated on synthetic complex pairs that differ in a target invariant. Yet, whether a model recovers that invariant or an artifact of the chosen discretization is seldom audited. To examine this, we design and test pairs of complexes that look identical to standard homological and local message-passing tests, have the same Hasse graphs, yet differ in the target invariant. We study two beyond-homology regimes: nonAbelian fundamental-group structure and homotopy-matched orientability. For the former, we construct genus-2 pairs that match homology, cell statistics, local incidence profiles, and rank-colored Hasse 1-WL behavior while differing in their fundamental groups. For the latter, we evaluate punctured torus-Klein bottle pairs and show that distance, boundary, cycle, and spectral signals can confound apparent orientability detection, motivating stronger matched controls. Under the tested settings, subcomplex-lifted architectures separate some pairs beyond the matched homological, local-incidence, and Hasse 1-WL controls, while a depth sweep reveals an empirical depth-dependent trend without establishing a general barrier or direct computation of the target invariants. These results show that claims of topological expressivity must be stated relative to the model representation, explicitly controlled information channels, and the precise scope of the benchmark.

0.23

Coresets for $k$-Means: When Does Uniform Sampling Suffice?

cs.DS cs.CG ·2026-10-22 ·Amir Carmel, Robert Krauthgamer
Abstract · 23 pages, 2 figures

Data summarization techniques, such as coresets, are essential for scaling clustering algorithms to massive datasets. We study uniform sampling, which is an extremely simple and ultra-efficient method to construct coresets, in the context of weak, stable, and strong coresets for the $k$-means problem in Hilbert space. We begin with the fundamental case $k=1$, and establish matching upper and lower bounds on the sample complexity of stable and weak coresets, finding that the two notions require different sample complexities. As a consequence, we obtain tight bounds for stable $1$-median coresets in $\ell_1$ and $\ell_2$ (and thus also Kendall--tau and Jaccard metrics), removing the dimension dependence of Carmel and Krauthgamer (ICLR 2026). Our analysis is substantially simpler than prior work and reduces to bounding the deviation of the sample mean. For the general $k$-means problem on $ρ$-separated instances, a well-studied structural assumption on the input, we show that uniform sampling yields a stable coreset for all $O(1)$-approximate solutions, and a strong coreset if the instance additionally has bounded kurtosis. In all these results, the sample complexity is independent of the dimension and the instance size.

0.23

A local-to-global principle for $f$-divergence homogenization

cs.IT math.PR ·2026-10-22 ·Aryeh Kontorovich
Abstract

We study the effect of replacing a family of independent statistical experiments by independent repetitions of their average. We give a simple, complete characterization of the $f$-divergences for which this operation increases divergence by at most a constant depending only on the generator, uniformly in the number of observations and the underlying measurable space. The characterization is governed by a local-to-global principle: a comparison restricted to two observations on a binary alphabet already determines the validity of the general inequality. It admits an explicit scalar formulation, which simplifies further for symmetric divergences. The result applies to arbitrary convex generators and requires no differentiability or symmetry assumptions. It recovers the homogenization inequality for total variation and identifies exactly the admissible members of the power-divergence family. In particular, whenever uniform homogenization fails, the obstruction is already present in a family of binary experiments with two observations. Moreover, the global comparison constant is controlled uniformly over generators with a common bound in the local inequality.

0.22

Identifying level-$k$ reasoning in repeated games: Strategies, beliefs, and cognitive ability

cs.GT econ.THecon.GN ·2026-10-22 ·David Gill, Yaroslav Rosokha
Abstract · 48 pages, including supplemental appendix. JEL: C73, D83, D91

In this paper, we identify level-$k$ reasoning in repeated games that operates at the level of a supergame strategy. First, we develop a model of level-$k$ reasoning that incorporates choices over strategies as well as beliefs about the strategies chosen by others. Then, jointly using elicited strategies and beliefs about strategies from the indefinitely repeated Prisoner's Dilemma, and applying a new notion of balance-optimal error tolerance, we classify a substantial fraction of subjects as level-1 or level-2. Moreover, we show that when level-$k$ reasoning operates at the level of a repeated-game strategy, cognitive ability and experience predict higher level reasoning.

0.22

The permutation and automorphism groups of univariate multiplicity codes

cs.IT ·2026-10-22 ·Eduardo Camps Moreno, Hiram H. López, Umberto Martínez-Peñas, Rodrigo San-José
Abstract

We give explicit characterizations of the permutations and linear isometries that leave (univariate) multiplicity codes invariant. We consider any evaluation subset of the finite field, any multiplicity and dimension, and the automorphisms are with respect to the Niederreiter-Rosenblom-Tsfasman (NRT) metric. For multiplicity one, we recover the permutation and automorphism groups of classical Reed--Solomon codes in the classical Hamming metric. In the main dimension range, automorphisms are uniquely determined by a projectivity that permutes the evaluation subset, up to multiplication by a nonzero scalar. For multiplicity one, permutations are given by affine maps that permute the evaluation subset. In contrast, for multiplicity at least two, permutations are only given by translations that permute the evaluation subset. Finally, we give an algebraic characterization of automorphisms of multiplicity codes for the folded Hamming metric.

0.21

Stochastic Geometry Meets Optimization: Data-Driven Cellular Planning

cs.IT cs.NI ·2026-10-22 ·Abhishek K. Shukla, Abhishek K. Gupta, Ketan Rajawat
Abstract

Deployment optimization of base stations (BSs) is critical for improving cellular network performance metrics such as coverage and capacity, impacting user experience, scalability, and infrastructure cost-effectiveness. Traditional methods that optimize precise BS locations do not generally provide broader insights into spatial deployment priorities or alternative viable deployments. Stochastic geometry approaches, on the other hand, offer a more interpretable intensity-based framework, but rarely utilize real-world user location data or investigate optimal BS deployment strategies. In this work, we propose a data-driven stochastic geometry-based approach that optimizes BS deployment intensity to maximize a network performance metric using samples from an unknown user-location distribution. We specialize the framework to an inhomogeneous Poisson point process (PPP) for cellular coverage analysis. The resulting infinite-dimensional problem is discretized over the deployment area and solved using sampled user locations. We characterize the discretization and sampling errors, apply stochastic mirror ascent to solve the empirical problem, and analyze the selection of concrete BS deployments using independent validation samples. Unlike traditional methods, our approach yields an optimized spatial intensity that reveals deployment priorities across regions and accommodates practical deployment constraints, multi-tier networks, and incremental deployment scenarios. Numerical experiments with synthetic and real-world spatial demand data demonstrate the proposed approach under different network configurations.

0.21

Fast Iterative Combinatorial Auctions via Deep Reinforcement Learning

cs.GT cs.LG ·2026-10-22 ·Luis F. Paul Tosta, Quinn Arnold, Gianluca Brero
theory
Abstract

Iterative combinatorial auctions are used in spectrum allocation, transportation, and industrial procurement, but rounds are costly. We introduce FastCombo, a deep reinforcement learning approach to sequential price discovery. Bayesian pricing has sharply reduced auction rounds even against subgradient benchmarks tuned with hindsight for each instance, but relies on a one-step objective and repeated online inference and optimization. We prove that greedily maximizing immediate clearing probability can require arbitrarily more rounds in expectation than an informative policy, even with the true valuation distribution known. FastCombo learns price policies offline, assigning credit to queries for the future decisions they enable and producing each online price update with a single neural-network forward pass. We formulate the objective of minimizing expected rounds to clearing in single-minded auctions as a partially observable Markov decision process and train a recurrent bidder-item graph policy over complete auction trajectories. In a one-item experiment, all five trained policies learn an informative opening, reducing expected rounds from 1.85 to 1.65 relative to the tested Bayesian one-step rule. Across four domains from the Combinatorial Auction Test Suite, FastCombo achieves competitive clearing while reducing measured online price-selection time per auction by a factor of roughly 300-900 relative to the standard Bayesian baseline. These results demonstrate the value of learning sequential price policies for both information acquisition and online computational efficiency.

0.21

Deterministic Constructions of Reed-Solomon Codes Beyond the Johnson Bound

cs.CC ·2026-10-22 ·Abhranil Chatterjee, Sumanta Ghosh, Zeyu Guo, Rohit Gurjar, Roshan Raj +1 more
Abstract

We give a deterministic quasipolynomial-time construction of Reed-Solomon codes that are list-decodable beyond the Johnson radius. Specifically, for every $ε\in(0,1]$ and all sufficiently large $n$, we construct an $[n,k]$ Reed-Solomon code of rate $R=Ω(ε/(1+\log(1/ε)))$ over a field of size $(nk)^{quasi-poly(nk)}$ that is $(1-ε,O(1/ε))$ list-decodable. The construction runs in quasi-poly(nk) time. We also give a deterministic construction of $[n,k]$-MDS$(3)$ Reed-Solomon codes for all integers $1\leq k<n$, with the same asymptotic bounds on field size and construction time. In the constant-rate regime, our field-size bound substantially improves the bound of $O(n^{k^{2k}})$ achieved by Roth's construction (Roth, IEEE TIT 2022). Our constructions exploit a connection to polynomial identity testing for read-once determinants, together with the quasipolynomial isolating-weight construction of Gurjar and Thierauf for linear matroid intersection (Gurjar and Thierauf, STOC 2017 and Computational Complexity 2020).

0.20

The Ternary and Quaternary Cases of the $\{2,4\}$-Distance Code Conjecture

cs.IT math.CO ·2026-10-22 ·Yen-Chi Roger Lin, Valjakas Djaljapayan, Wei-Hsuan Yu
theory
Abstract

For an integer $q\ge2$, let $A_q(n,\{2,4\})$ be the largest cardinality of an unrestricted $q$-ary code of length $n$ in which any two distinct words have Hamming distance $2$ or $4$. We prove that \[ A_q(n,\{2,4\})=1+\binom n2 \qquad(q\in\{3,4\},\ n\ge6), \] thereby settling the two remaining alphabet sizes in the conjecture of Boyvalenkov, Delchev, Zinoviev, and Zinoviev; the binary case was settled previously. The quaternary result implies the ternary result by alphabet inclusion. For $n\ge16$, the quaternary upper bound is proved combinatorially. After normalizing one word to zero, the weight-four supports form a $2$-intersecting family; exact support intersections force symbol synchronization and restrict support multiplicities. We combine the nontrivial intersection theorem with matching arguments and a three-coordinate projection lemma. The finite lengths $6\le n\le13$ are handled by a complete Hamming-isometry reduction to exact clique decisions, and $n=14,15$ by a smaller exact weighted-clique relaxation. All finite decisions are reproduced by two standalone integer-only C++17 verifiers provided as supplementary source files.

0.20

A Polynomial-Time Algorithm for $k$-Path Vertex Cover Reconfiguration on Trees under Token Sliding

cs.DS cs.DM ·2026-10-22 ·Duc A. Hoang
Abstract

A $k$-path vertex cover ($k$-PVC) of a graph is a set of vertices meeting every path on $k$ vertices. In Token Sliding, one token occupies each vertex of a cover; a move slides one token along an edge to an unoccupied vertex, preserving the cover property. We study whether two covers of a tree are joined by such moves. The case $k=2$ admits a linear-time decision algorithm. For every integer $k\geq2$, we give an $O(kn^2)$-time decision algorithm on trees of order $n$, with $k$ part of the input. Two covers are reachable precisely when their rigid tokens, which never move, occupy the same set and every component remaining after their deletion has the same token count. The proof transforms configurations of equal size without rigid tokens into a canonical configuration built on properly rooted subtrees, whose paths on $k$ vertices pass through their roots. A rooted-subtree recurrence identifies movable candidate tokens. We also give a direct recursive construction between the supplied endpoints, returning at most $8n^2\,3^{\lceil n/2\rceil}$ slides in $O(n^{10}3^{n/2})$ time. ChatGPT 6 Pro and Claude Fable 5.1 were used to assist with developing and reviewing mathematical arguments and drafting and revising the paper. The author takes full responsibility for the correctness and content of the paper.

0.20

Peak-Aware One-Way Trading: Competitive and Learning-Augmented Algorithms for Energy Procurement

cs.DS ·2026-10-22 ·Jackson Bibbens, Adam Lechowicz, Nicolas Christianson, Bo Sun, Mohammad Hajiesmaili
Abstract · 27 pages, 2 figures

We introduce peak-aware one-way trading ($\texttt{PA-OWT}$), an online problem in which a player must purchase a fixed quantity of an asset by a deadline, minimizing its total cost plus a charge proportional to the largest single purchase at any time. The model is motivated by electricity procurement, where the cost of energy rewards buying at low prices, but peak demand charges discourage concentrating purchases in short periods. Balancing these incentives is challenging when future prices are unknown: a large purchase can establish a peak that later decisions cannot undo. We develop a peak-aware online algorithm that balances price and peak control with deadline feasibility and achieves the order-optimal competitive ratio. We further extend our results to a learning-augmented algorithm that uses predictions of optimal purchasing decisions to achieve an order-optimal consistency-robustness tradeoff, exploiting accurate advice while retaining worst-case guarantees in the face of arbitrary prediction errors. Finally, we evaluate our algorithms in a case study on electricity procurement using historical electricity price traces to understand the benefits of peak-aware purchasing.

0.19

Equilibria and Learning in Competitive Market Making

cs.GT cs.LGq-fin.TR ·2026-10-22 ·Matteo Onger, Luigi Foscari
Abstract · 35 pages, 4 figures

We propose a market-making game played among $N \ge 2$ strategic market makers and a single non-strategic representative trader. In a continuous quote space, standard finite-game arguments do not apply, while non-convex and discontinuous payoffs rule out classical continuous existence results and complicate learning. Nonetheless, we characterize pure-strategy Nash equilibria (PSNEs) and show that at least one PSNE always exists under minimal assumptions on the underlying asset value distribution, and that the equilibrium bid-ask spread widens strictly with asset value volatility. To obtain a tractable learning framework, we discretize prices and, under an additional regularity condition, show that the time-averaged play of no-internal-regret learners approaches the continuous Nash price envelope as the tick size vanishes. Finally, we empirically validate our theoretical results through numerical simulations.

0.19

Randomization to Balance Performance under Nominal and Compromised Communication

cs.GT ·2026-10-22 ·Colton Hill, Brandon Collins, Philip N. Brown
Abstract

It is well-known that under nominal communication, where agents are able to share information with one another, game-theoretic methods provide system-level performance guarantees in multiagent cyber-physical human systems such as distributed resource allocation. However, approaches that depend on communication are frequently not robust when communication is compromised. While randomized decision-making can prevent arbitrarily poor outcomes under compromised communication, these performance guarantees often do not improve when communication abilities are restored. Thus, it remains unclear whether a single approach can perform well under nominal communication and remain robust under compromised communication. In this paper, we study whether a system operator can inform agents about how much they should randomize their choice when selecting actions in order to balance performance across both settings. We show that for any scenario (e.g., a multiagent resource allocation instance), there exists a level of randomization where nominal efficiency is near-optimal while compromised efficiency is strictly positive. We then show that this guarantee weakens as agents must choose between increasingly many resources. These results provide an initial step toward understanding randomization as a tool for robust performance across communication settings.

0.17

Hardness and Approximation for Compressing Deterministic Finite Automata

cs.DS ·2026-10-22 ·Philip Bille, Inge Li Gørtz, Christoffer Krogh, Christian Møller Mikkelstrup
Abstract · 18 pages, 4 figures. Has been submitted for review

The delayed deterministic finite automaton ($\text{D}^2\text{FA}$) is a powerful tool for compressing the prohibitively large deterministic finite automata (DFAs) used in modern signature-based deep packet inspection systems. A $\text{D}^2\text{FA}$ augments a DFA with unlabeled default transitions to remove transitions that are similar. This technique works well in practice and has inspired numerous subsequent studies. However, existing approaches do not guarantee a $\text{D}^2\text{FA}$ of optimal size as they restrict the default transitions to form a forest. In this paper, we consider the problem of finding an optimal $\text{D}^2\text{FA}$. We make the key observation that cycles of default transitions can be valid and produce a smaller $\text{D}^2\text{FA}$, provided that for every character in the alphabet, at least one state in the cycle has an outgoing transition labeled by that character. This observation raises fundamental questions concerning the computational complexity of finding an optimal $\text{D}^2\text{FA}$ and the approximation guarantee of the standard (forest-restricted) approach. We answer both questions by showing that it is NP-hard to find an optimal $\text{D}^2\text{FA}$ and by proving that the standard approach provides a $3/2$-approximation to the optimum. To establish NP-hardness, we construct a reduction from set cover using a DFA with gadgets where valid default transition choices correspond to selecting sets. To prove our approximation ratio, we compare the cycles in an optimal solution to a variant where the default transitions form a forest. Finally, we demonstrate that this $3/2$ ratio is tight via a constructed family of DFAs.

0.17

Iterative Randomized Rounding for $k$-Median in General and Euclidean Metrics

cs.DS ·2026-10-22 ·Jarosław Byrka, Yuhao Guo, Yang S. Hu, Shi Li, Yongho Shin +3 more
Abstract · 76 pages, 1 figure. The abstract was abridged to meet the arXiv requirement

We refine the iterative rounding framework of Byrka, Guo, Hu, Li, Wan, and Wang [FOCS'26], which gives a $(2+\varepsilon)$-approximation for $k$-median. Our algorithm opens $k+O(1)$ facilities with expected connection cost at most $1.9999$ times the standard LP value. Despite the small numerical gain, this shows that the LP integrality gap is below $2$ with a constant additive facility allowance, partially answering Li and Svensson's open question [SICOMP'16]. Their reduction yields a $1.9999$-approximation opening at most $k$ facilities. Our preprocessing makes the LP solution $\mathrm{poly}(\varepsilon)$-integral with a $(1+\varepsilon)$ loss in each client's expected connection cost and $O(1)$ extra facilities, without special treatment of clients with concentrated fractional facilities. We replace batch processing with single-facility openings and combine a tighter analysis for clients with unequal distances to their facilities with a complementary algorithm for nearly uniform distances. In Euclidean metrics, we obtain a $1.71955$-approximation for $k$-median and a $1.4561$-approximation for uncapacitated facility location (UFL). For every $p\ge2$ and sufficiently small $\varepsilon>0$, we obtain a $((3^p+2)/3+\varepsilon)$-approximation for $k$-clustering with $p$-th power distance costs. For $p=2$, our $(11/3+\varepsilon)$-approximation for $k$-means improves the $(3+\ln2+\varepsilon)$ bound of Anand, Charikar, Cohen-Addad, Gao, Grandoni, Lee, Sharma, and van Wijland [SODA'27]. The Euclidean $k$-median, $k$-means, and UFL ratios are below the respective general-metric hardness bounds $1+2/e$, $1+8/e$ [STOC'02], and $1.463$ [J. Algorithms'99]. The Euclidean $k$-median bound relies on a geometric inequality proved via a nonnegativity certificate constructed from positive-semidefinite kernels.

0.14

Balanced Posted Prices in Stationary Two-Sided Markets

cs.GT ·2026-10-22 ·Ilan Reuven Cohen
benchmark
Abstract · 30 pages

We study an intermediary that buys nonperishing items from arriving sellers and sells them to arriving buyers, subject to limited storage. Sellers and buyers arrive according to independent Poisson processes. The intermediary pays and charges the same fixed item prices, aiming to maximize long-run gains from trade: buyer value minus seller cost. Our benchmark may choose allocations directly, has unlimited storage, and must purchase each unit before allocating it. Its optimal long-run gain rate equals the value of a linear program of average purchase and allocation rates, with or without knowledge of future arrivals. For buyers whose bundle values are maxima of nonnegative additive valuations (XOS), we compute static common prices that guarantee half of this benchmark for arbitrary finite item capacities, even one storage slot per item, regardless of how ties are resolved without using future arrivals. The prices are computable in polynomial time using demand and supporting-value queries. The construction equalizes itemwise lower bounds on buyer and seller contributions to gains from trade, accounting for procurement costs. We call this contribution balance. As a single-item foundation, we also equalize the rates of purchases and sales selected by the price, before checking inventory availability. We call this rate balance. For capacity $C$, it yields the tight guarantee $C/(C+1)$. The two principles parallel proofs of the classical prophet inequality based on half the expected benchmark and on a median threshold, adapting these balancing ideas to recurring two-sided markets with limited inventory.

0.13

Classification of binary, ternary and quaternary Hermitian LCD codes

cs.IT ·2026-10-22 ·Keita Ishizuka
Abstract

We extend the complete classifications of linear complementary dual (LCD) codes from length 15 to length 19 in the binary case, from length 10 to length 15 in the ternary case and from length 12 to length 13 in the quaternary Hermitian case. The classification rests on a reduction of the monomial equivalence of LCD codes to graph isomorphism on the coordinates, combined with canonical augmentation.

0.13

Closing All-Exp Gaps in Local Search for the Traveling Salesman Problem and Maximum Constraint Assignment

cs.DS math.CO ·2026-10-22 ·Sophia Heimann, Hung P. Hoang, Stefan Hougardy
theory
Abstract · 37 pages

We prove the all-exp property for the Traveling Salesman Problem with the $2$-opt neighborhood, and for $(3,2,2)$- and $(2,3,2)$-Maximum Constraint Assignment with the flip neighborhood. The all-exp property states that infinitely many instances have an initial solution from which every sequence of improving moves to a local optimum, and hence every run of local search, has exponential length. For the Traveling Salesman Problem, the all-exp property was previously known for every $k$-opt neighborhood with $k \ge 3$, so our result for $2$-opt completes the picture for all $k$-kopt neighborhoods. In $(p,q,r)$-Maximum Constraint Assignment, each constraint involves at most $p$ variables, each variable occurs in at most $q$ constraints, and each domain has size $r$. Decreasing any of the parameters of $(3,2,2)$ or $(2,3,2)$ yields a problem in which every sequence of improving flips has polynomial length, so both Maximum Constraint Assignment results are tight. All three results settle long-standing open problems. Our proofs build on a common construction idea, which reveals a close connection between the local-search dynamics of the three problems.

0.11

One-shot entanglement generation and distillation via Petz recovery and binary hypothesis testing

cs.IT quant-ph ·2026-10-22 ·Matthew Can, Mark M. Wilde
Abstract · 53 pages, 6 figures

How much entanglement can two distant laboratories share when their connection is noisy? We relate this question to deciding which of two quantum hypotheses is present. Our one-shot bounds apply to entanglement generation over a quantum channel and to one-way entanglement distillation from a shared state. We use Cheng's trace inequality to bound the infidelity of a Petz decoder in terms of a binary hypothesis testing cost. Applying the inequality before averaging over codes reduces the error analysis to computing a second moment of random subspaces. The proof covers every integer target dimension and treats rank-deficient states explicitly. A corollary expresses the achievable entanglement through hypothesis-testing coherent information, with a single logarithmic error slack and exact integer rounding. We also prove an error-preserving conversion from entanglement generation codes to entanglement transmission codes: the target dimension decreases by only a constant factor at fixed error, and the decoder is the original decoder followed by local processing. Consequently, for both entanglement generation and entanglement transmission, arbitrary pure reference-input states yield the same coherent-information rate, second-order Gaussian correction at the original error tolerance, and explicit achievable logarithmic remainder. For commuting testing hypotheses, the logarithmic remainder improves to a constant. These results give a direct route from binary distinguishability to explicit recovery guarantees and finite-resource quantum communication rates.

0.11

Improved Bounds for the Entropic Central Limit Theorem in the Log-Concave Case

cs.IT math.PR ·2026-10-22 ·Tom Maitre
theory
Abstract

Let $μ$ be a log-concave isotropic probability measure on $R^d$ , and let $μ^n$ denote the law of the normalized sum of $n$ independent random vectors with law $μ$. We prove that \[ H(μ^n|γ)\leq \frac{C}{n} \min\left\{1+\log(ed)^2H(μ|γ)\ ,\ d\right\}. \] Compared to the previous best known bound, our estimate replaces the polynomial dependence on the dimension by a polylogarithmic one. The proof relies on a new combination of the De Bruijn identity with Eldan's stochastic localization process.

0.11

Assortment optimization under the perception-adjusted Luce model

cs.GT math.OC ·2026-10-22 ·Gerardo Berbeglia, Mashbat Suzuki, Matthew Tam
theory
Abstract · 17 pages, 1 figure

The Perception-Adjusted Luce Model (PALM) introduced by by Echenique et al. (2018) is a sequential choice model that can accommodate violations of regularity and choice overload, behaviors not captured by the utility-maximization framework. Flores et al. (2019) studied the assortment problem under the PALM and proved that, for the case of two levels, there is always an optimal solution that is revenue-ordered by level. They also conjectured that revenue-ordered by levels remain optimal for the general PALM with any number of levels. We resolve this question in the affirmative by showing that every PALM instance admits an optimal assortment that includes, within each level, all products at or above a level-specific revenue threshold. The proof involves first showing how the conjecture which deals with a discrete optimization problem, can be reduced to proving certain property of a related continuous optimization problem. Specifically, we study a continuous refinement of the assortment decision under the same PALM choice rule. Although the refinement permits partial utility selections, we prove it always has an optimum that selects complete revenue tiers. This allows us to show that the assortment problem is polynomial-time solvable when the number of levels is fixed.

0.10

Optimal Mixing for Local Edge-Coloring Dynamics on Trees

cs.DS math.PR ·2026-10-22 ·Haoran wang
Abstract

We study the mixing time of two local Markov chains for uniformly sampling proper edge colorings of trees. For a tree with $n$ vertices and maximum degree $Δ$, we establish optimal mixing-time bounds for the single-edge Glauber dynamics when $q\geΔ+2$ and for the neighboring-edge heat-bath dynamics when $q\geΔ+1$. Both chains mix in $O_Δ(n(\log n+\log(1/\varepsilon)))$ steps, uniformly over the number of colors $q$. These bounds are tight in their dependence on $n$ for fixed $Δ$. Our proof establishes log-Sobolev inequalities through local color release, reusable recoloring buffers, and component entropy mixing. The two dynamics require different local constructions: a two-spare-color buffer for single-edge updates and a second-layer controller for neighboring-edge updates. The resulting entropy analysis applies to general bounded-degree trees, including regimes where worst-case spatial mixing fails.

0.08

A Weakly Group Strategy-Proof and Pareto-Stable Matching Mechanism with Two-Sided Indifferences under Gross Substitutes

cs.GT econ.TH ·2026-10-22 ·Yasushi Kawase, Ryosuke Sato
Abstract

We study finite many-to-one matching markets with indifferences on both sides and publicly known firm preferences over sets of workers. A standard approach breaks ties uniformly at random and runs deferred acceptance, which is strategy-proof for workers but may select a weakly stable matching that is Pareto dominated under the agents' original preferences. For the broad class of markets in which each firm's preferences admit a gross substitutes (GS) representation, we construct a deterministic generalized deferred acceptance mechanism that is weakly stable, Pareto efficient with respect to both sides, worker weakly group strategy-proof, and implementable in strongly polynomial time given value oracles for the GS representations. This domain includes nonresponsive firm preferences, such as additive scores under matroid constraints and assignment values. Our analysis bridges ordinal matching with indifferences and competitive equilibria under imperfectly transferable utility. Although the mechanism uses only ordinal comparisons, it returns a seller-optimal equilibrium matching of a tiered auxiliary market, and the equilibrium structure of this market yields the stability, efficiency, and incentive guarantees.

0.07

An Improved Bound for Testing $k$-Colorability

cs.DM ·2026-10-22 ·Bangjie Xu
theory
Abstract

We study the sample complexity of testing $k$-colorability in the dense graph model. A graph is $\varepsilon$-far from being $k$-colorable if at least $\varepsilon n^2$ edge deletions are required to make it $k$-colorable. We prove that a uniformly random induced subgraph on $O(k\log k/\varepsilon)$ vertices is non-$k$-colorable with probability at least $2/3$ whenever the original graph is $\varepsilon$-far from $k$-colorable. The proof combines a palette-based coloring process with a container-style encoding-and-reconstruction argument. This yields $O(k\log k/\varepsilon)$, which is no worse than the previous state of the art $O(k\log (1/\varepsilon)/\varepsilon)$ throughout the nonvacuous parameter range, and is an improvement when $\log k=o(\log(1/\varepsilon))$. For each fixed $k$, combining this upper bound with the known $Ω(\log k/\varepsilon)$ lower bound gives $Θ(1/\varepsilon)$ dependence on $\varepsilon$.

0.06

Robotics, Control & Signals 106

CMAR-search: Commonsense and Memory Augmented Reasoning for Object Search in Dynamic Interactive Environments

cs.RO ·2026-10-22 ·Kaiyao Liao, Qingfeng Li, Xinlei Zhang, Chen Chen, Qing Sun +1 more
top venue
Abstract · Published in ICRA 2026. 8 pages, 5 figures

Dynamic interactive object search in large-scale human environments presents substantial challenges for existing methods. Current scene representations like 3D Scene Graphs (3DSG) only provide coarse-grained spatial segmentation and cannot identify functional areas such as storage or leisure areas. Without functional area understanding, existing methods are constrained to exhaustive sequential exploration at large scales, resulting in inefficient search behaviors -- particularly in open-layout environments with numerous interactive objects such as drawers and cabinets. Moreover, these methods lack adaptability to environmental dynamics such as object relocations. To address these limitations, this paper proposes CMAR-search, a novel framework built upon Commonsense and Memory Augmented Reasoning (CMAR). Our approach leverages commonsense about area functionalities and aggregates environmental memory to construct a Functional 3D Scene Graph (F3DSG), which organizes the environment into functional areas with their associated containers. Through this structured representation, CMAR enables hierarchical action planning at both macro-area and micro-container levels, empowering the system to efficiently identify and inspect semantically relevant areas for effective object search. Notably, CMAR continuously integrates real-time perception, accumulated memory, and commonsense to dynamically relocalize objects in changing environments. Extensive experiments in simulation and real-world settings demonstrate that CMAR-search significantly surpasses state-of-the-art baselines in both success rate and search efficiency for object search in dynamic interactive environments.

0.61

OASIS: Online Anchor-centric Scene Inference System for Open-Vocabulary Retrieval and Navigation in Dynamic Environments

cs.RO ·2026-10-22 ·Kaiyao Liao, Qingfeng Li, Chen Chen, Xiaoze Wu, Qing Sun +1 more
top venue
Abstract · Accepted for oral presentation at ACM Multimedia (ACM MM) 2026. 17 pages, 10 figures, including supplementary material

Reliable robotic mapping in dynamic real-world environments is essential for embodied intelligence and long-term autonomy. However, existing methods often involve a trade-off: lightweight online mapping is efficient but captures limited semantic context, while dense mapping provides richer semantics at higher computational cost and is more vulnerable to distractors. They also struggle to relocalize objects after scene changes. To address these challenges, we propose the Online Anchor-centric Scene Inference System (OASIS), a unified system for online mapping, complex open-vocabulary retrieval, and navigation in dynamic environments. Motivated by the observation that indoor scenes are organized around stable semantic anchors, OASIS represents the environment using a sparse set of anchor nodes, each maintaining a unified multimodal representation that integrates semantic descriptions, visual features, and geometric states for reasoning, verification, and navigation. Its compact anchor-centric memory supports on-demand evidence extraction and hierarchical retrieval, reducing redundant storage while improving retrieval reliability. To handle scene dynamics, OASIS further reuses transient observations collected during navigation to relocalize moved objects. Experiments in both simulation and real-world environments, including standard retrieval benchmarks and navigation tasks under both static and dynamic settings, demonstrate that our system can perform complex open-vocabulary retrieval and navigation, while also enabling relocation of displaced objects after scene changes.

0.59

FLORA: An Open Model Suite for Scaling Robotics Data Engines

cs.RO ·2026-10-22 ·Sicong Leng, Yunze Man, Zhiqi Li, Jing Wang, Yuhe Zhang +12 more
benchmark
Abstract

The life-cycle of robotics foundation models crucially depends on data labels. Outcome reward labels are needed for monitoring real-world deployments. Process reward labels are needed during RL post-training. Timestamped captions are needed for filtering and balancing pretraining data. We present Flora, an industrial-quality data engine used for training GR00T models, that simultaneously covers these capabilities. Flora achieves SOTA scores on the captioning benchmark WGO-Bench and reward benchmarks RoboReward and Robometer. On internal benchmarks it is competitive with GPT-6 Astra while beating other frontier models. The model is substantially cheaper than API-based alternatives, roughly 125 times cheaper than GPT-6 Astra. Florais battle tested at scale, labeling 200,000 hours of data per day across 512 GB300s. By open-sourcing our data engine, we enable researchers to scale custom robotics datasets and explore emerging applications such as physical auto-research and human-to-robot transfer.

0.57

Memory before Goal: Role of Pre-History Memory for Humanoid Loco-Manipulation

cs.RO cs.CVcs.LG ·2026-10-22 ·Chenyuan Chen, Ran Ding, Lechen Zhang, Zewen He, Yoshihiko Nakamura
benchmarkcode/data
Abstract · 8 pages, 4 figures. Project page: https://memory-before-goal.github.io. GitHub: https://github.com/crystalccy1/MemoryBeforeGoal

Humanoid loco-manipulation in dynamic scenes is non-Markovian: cues vanish as objects change or are hidden, or leave the view as the robot turns and walks, yet whole-body vision-language-action (VLA) policies see only the present. We present WholeBodyMem, an object-guided perceptual memory that keeps the pre-history of object events seen before the robot acts on its goal. SAM3 tracks the task objects, and their events causally select key frames within an adaptive budget, which the action expert reads by cross-attention together with object tokens encoding appearance, geometry and time. We also contribute HuMeGo, a humanoid memory benchmark with six simulated and six real-world Unitree G1 tasks. Keeping the right frames rather than more frames, WholeBodyMem surpasses memoryless and uniform-memory baselines, raising its base policy's success from 29.1% to 72.6% in simulation and from 36.7% to 88.3% on the real robot, while also improving the standard SIMPLE tasks (70.2% to 86.9%).

0.53

To Route or to Adapt? Multi-Expert Diffusion Planners in Unseen Regions

cs.RO ·2026-10-22 ·Yun Li, Ehsan Javanmardi, Simon Thompson, Jiachen Gong, Qunli Zhang +5 more
publishedcode/data
Abstract · Accepted to the 10th Conference on Robot Learning (CoRL 2026). 36 pages. Code and weights: https://github.com/tlab-wide/RouteOrAdapt

Diffusion trajectory planners with domain-specific experts face a practical question in unseen regions: can they rely on selecting among existing experts, or must the planner itself adapt? We study this question using a planner with a shared adapter and per-domain expert adapters. Across three real-world held-out maps, none of eight test-time routing schemes improves the fleet-mean lateral offset over the static default, which uses the same expert for every scene. Even an oracle that selects the best expert for each scene reduces the mean offset by only 0.9 m. In contrast, reward-based fine-tuning on just 20 target-region scenes yields substantially larger gains. On matched evaluation slices, joint replay achieves a fleet-mean offset of 6.9 m, compared with 12.4 m for the static default. Comparisons under the same target-region training budget show that target-region training drives the improvement, while replaying training scenes limits forgetting on the original ODDs. On nuPlan, fine-tuning across training cities also improves open-loop offsets, although closed-loop performance deteriorates. Further diagnostics reveal that two test-time adaptation baselines achieve low offsets by producing degenerate trajectories. These results show the limited headroom for selecting among existing experts and the greater open-loop gains available from adapting the planner.

0.48

Stochastic Geometry Analysis and Optimization for Multi-LEO Satellite-Assisted Communications

eess.SP ·2026-10-22 ·Xu Hu, Bin Lin, Ping Wang, Xiao Lu, Weihua Zhuang +1 more
published
Abstract

Low Earth orbit (LEO) satellites are vital in providing seamless connectivities to ground users (GUs). However, complex satellite-to-ground propagation and non-negligible satellite failure probabilities make single-satellite transmission unreliable. To address this issue, we consider a multi-LEO-satellite communication network (mLEO-CN), where a typical GU is served by multiple LEO satellites. The satellite locations are modeled as a binomial point process on a sphere and ordered according to their distances to the GU. To improve communication reliability, a serial communication scheme is adopted, in which the subsequent satellite is activated only when all preceding satellites fail to satisfy the target signal-to-noise ratio threshold. By employing the tool of stochastic geometry, we derive key performance metrics, including overall transmission success probability, average transmission rate, and end-to-end delay under the serial communication scheme by incorporating both shadowed Rician (SR) fading and satellite failure probabilities. To improve analytical tractability, we derive asymptotic expressions for the end-to-end transmission success probability in two special cases by leveraging the asymptotic expression of the SR fading model and a Gamma-distribution, and further propose a cluster-based approximation approach for the overall transmission success probability. Building on these analyses, we formulate an optimization problem to maximize the ratio of the overall transmission success probability to the end-to-end delay, and obtain its solution using a one-dimensional search approach. Numerical results validate the analysis, reveal the impact of key parameters, and identify the optimal number of serving LEO satellites under different constellation altitudes.

0.46

Stability-Preserving Reinforcement Learning for Heterogeneity-Aware Composite Max-Pressure Traffic Signal Control

eess.SY ·2026-10-22 ·Jichen Zhu, Chaopeng Tan
Abstract

Reinforcement learning (RL) provides a flexible mechanism for adapting traffic signal control to heterogeneous traffic and sensing conditions, but directly learned signal actions do not automatically inherit the analytical properties of structured traffic-control methods. This paper proposes Com-MP, a theory-guided RL framework that learns a context-dependent composition of multiple max-pressure (MP) control preferences while retaining decentralized MP phase selection. The learned policy adapts the relative influence of constituent MP formulations according to local traffic and sensing conditions, allowing data-driven adaptation to operate within a structured traffic-control framework. We show that such intersection-specific and time-varying composition preserves strong queue stability for every demand in the strictly stabilizable region under any admissible non-anticipative composition policy, thereby separating the performance-adaptation role of learning from the stability property of the underlying controller. In a representative implementation combining detector-supported queue-based MP and connected-vehicle-supported travel-time MP, experiments on a real-world 45-intersection network show that Com-MP reduces average delay by 15.6\% and 8.6\% relative to the two constituent controllers, respectively, and by about 7.0\% relative to the best fixed composition. The learned composition responds systematically to local traffic and sensing conditions, remains effective under heterogeneous and degraded sensing, and can be transferred without retraining to an unseen network with an average delay only 2.4\% higher than that of a policy trained directly on the target network. These results show how learning-based adaptation can be introduced within a stability-preserving traffic-control structure, providing a theory-guided pathway toward trustworthy learning-enabled traffic signal control.

0.46

Extending Harmonic Function Methods via Root Locus for Coordinated Path Generation

cs.RO eess.SY ·2026-10-22 ·Matteo Frascarelli, Lucia Pallottino, Alberto Landi
code/data
Abstract · 16 pages, 6 figures. Extended version of a paper submitted to Automatica. Corresponding code: https://github.com/mfrascarelli/grl-certified-…

We consider the offline generation of coordinated paths for N interchangeable robots and N goals in planar environments with polygonal obstacles. After a conformal map to a circular domain, starts and goals become the poles and zeros of a rational function R, and the paths are the branches of its Generalized Root Locus (GRL), i.e. the streamlines of the logarithmic potential -log R. We show that, for every phase outside a finite set of critical phases and a set of boundary phases, the induced gradient motion drives each robot to a distinct goal along pairwise disjoint paths that never reach the obstacles; the robot-goal assignment is produced by the construction, and with the gain used as a clock all robots move as the closed-loop roots of the GRL and arrive together at a prescribed time. Obstacles become streamlines through successive Milne-Thomson reflections, and for a finite number of reflections a finite test certifies safety for the potential that is actually computed. Reciprocal root-locus planning is recovered as a special case. The method is illustrated on two polygonal workspaces, 100 random instances compared with independent harmonic planning and artificial potentials, and a scene with ten robots and ten obstacles.

0.46

EgoForesight: Effectively Scaling Policy Learning with Noisy Egocentric Actions

cs.RO cs.CV ·2026-10-22 ·Chubin Zhang, Changyuan Wang, Yirui Wang, Ao Li, Shilin Ma +5 more
code/data
Abstract · Project page: https://lin-shan.com/EgoForesight

Scaling policy learning requires broad experience and reliable supervision. Robot demonstrations provide high-quality trajectory supervision but are costly to collect. Egocentric videos broaden task and environment coverage, yet motion reconstruction and retargeting introduce noisy action labels. To address this challenge, we introduce EgoForesight, a vision-language-action framework that learns jointly from both sources by reducing the influence of unreliable action labels and exploiting demonstrated outcomes as complementary supervision. A joint vision-action flow couples future visual representation prediction with action generation, unifying reliability-aware action learning with outcome-informed representation transfer. Specifically, reliability-conditioned corruption allocates action supervision across noise levels according to source reliability, while reliability-weighted regression regulates each source's contribution. Complementing this treatment of noisy labels, privileged-future representation distillation draws supervision from demonstrated outcomes. An exponential-moving-average teacher observes the current scene and clean future to encode action-relevant structure from the resulting scene change. Its deeper representations supervise an earlier student layer receiving a corrupted future, providing outcome-informed structure beyond noisy action annotations. Together, these mechanisms support effective scaling with noisy egocentric actions. Over the tested range, EgoForesight exhibits steeper scaling slopes than the baseline, reducing real-world open-loop action MSE by 20.5% and improving RoboCasa few-shot success by 8.1 percentage points at 30,000 hours of pretraining data. After post-training, it also achieves strong absolute performance in both simulated and real-world dexterous manipulation.

0.45

SCOPE: Scene-graph Construction Optimized by Plan Error-correction for Robot Long-term Planning Datasets

cs.RO ·2026-10-22 ·Chaoyue Li, Zeyu Gao, Yanfeng Lu, Fan Luo
publishedcode/data
Abstract · 8 pages. Accepted at IROS 2026. Project page: https://scope-page.github.io/

Task and Motion Planning (TAMP) stands as a fundamental challenge in robotics, essential for enabling autonomous agents to execute complex, long-horizon tasks in diverse environments. While Large Language Models (LLMs) introduce a new paradigm for TAMP by leveraging extensive common sense, their inherent reasoning limitations hinder direct application, and fine-tuning strategies to enhance these capabilities are currently bottlenecked by the scarcity of high-quality datasets. Existing automated generation methods often suffer from homogeneity, limiting model generalization, and unsolvability, where hallucinated scene graphs create physical mismatches with task goals. To address these issues, we propose Scene-graph Construction Optimized by Plan Error-correction (SCOPE), an automated framework designed to generate diverse and executable TAMP datasets. SCOPE incorporates multi-level diversity metrics to maximize the information entropy of scenarios. Crucially, it introduces a Plan Feedback Error-correction pipeline that utilizes planner execution failures as feedback signals to retroactively rectify logical inconsistencies in scene graphs, ensuring physical feasibility. Experimental results demonstrate that SCOPE exhibits greater robustness to erroneous scene graphs and generates higher-quality data. When utilized for LLM fine-tuning, SCOPE-generated data yield a downstream TAMP model that achieves an absolute 12.33% gain in success rate over its counterpart fine-tuned with data generated by DELTA, a state-of-the-art PDDL-based planner.

0.45

Resource Slicing for Time-Sensitive Data Traffic in Non-Terrestrial Networks with FSO ISLs

eess.SP ·2026-10-22 ·Abid Afridi, Alexis A. Dowhuszko, Risto Wichman, Jyri Hämäläinen
published
Abstract · 6 pages, 7 figures, 1 table. Accepted for presentation at IEEE GLOBECOM 2026 Workshops

In multi-hop Low Earth Orbit (LEO) satellite networks, time-sensitive packets experience queueing across multiple satellite hops, causing per-hop waiting times that accumulate and make the maximum end-to-end (E2E) delay difficult to guarantee. This paper presents a novel multi-hop slicing framework that reserves time-frequency resources for each time-sensitive flow of data originating at each satellite, aiming to maintain the reservation of communication resources along every inter-satellite link (ISL) on its route to the ground station (GS). The number of time-frequency resource blocks (RBs) reserved per-hop is determined from the data packet arrival statistics, E2E delay budget, and target level of reliability. Reservations on consecutive links are temporally aligned to enable pipelined forwarding of the time-sensitive data flows aggregated from terrestrial user terminals on the different LEO satellites. We evaluate the proposed framework against conventional scheduling and resource-allocation baselines, including pure time-division multiple access (Pure-TDMA), first-in-first-out (FIFO), and farthest-satellite-first (FSF), over the same multi-hop LEO satellite path. Under light traffic conditions, the baseline schemes provide comparable delay performance; however, their high-percentile delays increase sharply as data traffic load grows, whereas the proposed framework exhibits substantially slower delay growth. At the high arrival rate and high reliability target considered, the proposed scheme requires about 31 ms, a factor of 2.5-6.8 lower than the 76, 180, and 210 ms required by FIFO, FSF, and Pure-TDMA.

0.43

Archive coverage constrains abstaining retrieval for transient stability screening more than conditional accuracy

eess.SY ·2026-10-22 ·Gregor Cerar, Denis Sodin, Matjaž Škrlec, Urban Rudež, Mihael Mohorčič +1 more
code/data
Abstract · 34 pages, 4 figures, 9 tables, including Supplementary Information. Submitted to Scientific Reports. Code and BUS39 data: https://github.com…

Power-grid transient stability is assessed by simulating credible faults, but repeating simulations as operating conditions change is costly, motivating faster critical clearing time estimators. Within the AI-ASSIST project, we developed a retrieval-based estimator around three adopted requirements: estimates retain their supporting states, retrieval abstains without compatible support, and support statistics are tested against observed error. Abstention turns estimator uncertainty into a demand on the archive. Exact switching-configuration equality is used. The dense synthetic BUS39 archive covers 99.9996% of 1,044,797 records; the operational ELES archive covers 73.47% of 322,823 records. At query time, 59.37% of ELES states have an earlier same-topology candidate. Under causal availability, retrieval covers 56.34% of records; on records covered under both protocols, mean absolute error increases by 10.0%. No tested support diagnostic ranks first on both grids. Most tested supervised regressors have lower error than retrieval on BUS39, all higher on the matched retrieval-covered ELES population. On ELES, gradient boosting becomes more accurate than retrieval when temporally adjacent states are excluded, so the accuracy trade depends on the archive. Under the adopted compatibility key, archive availability constrains deployment more than conditional accuracy on ELES. Compatible coverage and chronological availability are auditable before an estimator is fitted.

0.38

Geometry-Dependent Properties of Physically Crosslinked PPy-PVA Hydrogels Enable Tunable Iontophoretic Drug Delivery

cs.RO physics.med-ph ·2026-10-22 ·Heather Jin Hee Kim, Mohsen Alishahi, Kyuin Park, Mikaila K. Roncevich, Larissa M. Shepherd +1 more
top venueunusual
Abstract

Objective: Iontophoretic drug delivery has primarily relied on current modulation to regulate drug transport. Here, we investigate hydrogel geometry, microstructure, and local electric fields as alternative mechanisms for controlling transdermal delivery. Methods: Bovine serum albumin (BSA) was encapsulated in polypyrrole-polyvinyl alcohol (PPy-PVA) hydrogels. Five geometries, including Hilbert curve-based designs, were fabricated to vary pore architecture and local electric field distributions. We optimized PPy content, drug loading, and current density. Electric fields were evaluated using finite element analysis, and microstructures were characterized by scanning electron microscopy. Drug transport was assessed using Franz diffusion cells with artificial membranes. Results: Hydrogel geometries exhibited distinct microstructures and electric field distributions. Iontophoresis substantially enhanced permeability compared with passive diffusion, while individual geometries produced different cumulative release, flux, and permeability profiles. The average iontophoretic permeability was (3.80 +/- 1.93) x 10^-5, exceeding previously reported values using artificial membranes. Conclusion: Geometry-driven fabrication of drug-loaded hydrogels enables tunable transport characteristics under both iontophoretic and passive conditions. Significance: Geometry-mediated control offers an alternative approach to regulating iontophoretic drug delivery without requiring individual current-density modulation.

0.38

Demonstration-Based Robotic Feeding Using Dynamic Movement Primitives

cs.RO ·2026-10-22 ·Kenzhi Iskandar Wong, Kitti Thamrongaphichartkul, Sri Harsha Turlapati, Yanshuo Lu, Kanichi Doi +8 more
published
Abstract · Accepted on RAAI 2026

This paper presents a demonstration-based robotic feeding framework that integrates human demonstration, Dynamic Movement Primitive-based motion planning, vision-based scene perception, and robot control. A human operator first demonstrates the scooping and feeding motions while the position and Euler-angle orientation of the spoon are recorded using a motion-capture system. The recorded trajectory is segmented into scooping and feeding motions, which are encoded using separate Dynamic Movement Primitives. Before task execution, an RGB-D camera estimates the participant's mouth position, while two AprilTags are used for robot-frame registration and tray localization. The detected tray pose is used to position the scooping motion, whereas the detected mouth position defines the positional goal of the feeding motion. The approach, scooping, and feeding trajectories are then executed sequentially by a Kinova Gen3 manipulator through a ROS~2-based framework. A real-world experiment demonstrates the feasibility of the integrated pipeline and shows that the learned spoon motions can be reproduced within the detected experimental scene.

0.38

Skeleton-Grounded Latent Action Model for Robot World Models

cs.RO ·2026-10-22 ·Qi Lv, Jiayi Lin, Yuxin Jiang, Jianzong Wang, Xiang Deng +1 more
code/data
Abstract · Project Page: https://aopolin-lv.github.io/SkeLAM/

Robot world models simulate how environments evolve in response to robot actions. Latent action models enable learning dynamics from videos by encoding visual transitions as proxy actions. However, reconstruction-driven latent actions mix robot motion with object responses and other scene changes, complicating their correspondence with robot controls. We propose SkeLAM, a skeleton-guided latent action model for controllable robot world modeling. SkeLAM decomposes latent actions into a kinematic component grounded by robot skeleton supervision and a residual component retaining complementary visual dynamics. Cross-modal swap reconstruction encourages RGB-derived and skeleton-derived kinematic representations to be interchangeable during video reconstruction. The world model incorporates both components through separate conditioning pathways, while branch dropout reduces its dependence on future-derived residual information. A lightweight adapter maps robot actions to the kinematic space, and subsequent world-model fine-tuning adapts generation to these predicted conditions. At inference, the model generates future observations from the current observation and robot actions. On DROID, SkeLAM reduces end-effector ADE by 19.5% and LPIPS by 11.1% compared with DreamDojo. Motion probing supports kinematic specialization, while qualitative results demonstrate zero-shot human-to-robot motion transfer in generated rollouts. Policy evaluation shows agreement between predicted success rates and those measured in real robot executions.

0.38

Telling Neural Network What We Know: Weight Configuration for RNN and Its Application to NextG Real-Time Neural Receiver

eess.SP ·2026-10-22 ·Ummay Sumaya Khan, Lizhong Zheng, Lingjia Liu
Abstract

AI-enabled receivers offer a promising approach to adaptive physical-layer processing, but conventional neural receivers often require extensive offline training and computationally expensive backpropagation, limiting their suitability for low-latency cellular systems. Moreover, classical MIMO-OFDM receivers rely on accurate channel estimation, leading to substantial pilot overhead. This paper introduces skip-RNN, a lightweight skip-augmented recurrent neural receiver that performs online time-domain receive processing using fixed recurrent and skip branches with an adaptive linear readout. We show that the skip-RNN can be interpreted as a structured deconvolution filter whose recurrent and skip branches realize IIR and FIR basis functions, respectively. We develop a domain-knowledge informed weight-configuration framework for the skip-RNN by exploiting its structural correspondence with the target stable and causal inverse filter of mixed-phase channels. The dominant IIR structure extracted from estimated channel statistics is used to configure the recurrent dynamics, while the non-minimum-phase component is addressed through the skip branch and a statistical FIR readout prior incorporated into prior-centered regularized online readout adaptation. We further derive an approximation-error bound that characterizes the resulting configured neural-receiver representation. Simulations over standardized 3GPP channels demonstrate the effectiveness of the configured neural receiver under extremely low pilot overhead.

0.37

FocusVLA: Future-Intent Guided Dynamic View Routing for Vision-Language-Action Models

cs.RO cs.AI ·2026-10-22 ·Huixin Xu, Ziya Yu, Peng Li
published
Abstract · Accepted at CoRL 2026

Vision-language-action (VLA) models have become an important paradigm for robotic manipulation by unifying visual perception, language understanding, and action generation. Since reliable manipulation requires sufficient observation of objects, end-effectors, and their spatial relationships, multi-view inputs are widely used to provide complementary spatial information. However, our experiments show that existing VLA models do not fully exploit such multi-view complementarity. The policy tends to persistently rely on the wrist view, lacking the ability to dynamically adjust global-local view dependency across manipulation stages, which limits stage adaptability and robustness. To address this issue, we propose FocusVLA, a stage-aware multi-view routing framework. FocusVLA predicts future interaction intent from historical actions and uses it to guide dynamic view routing during action generation. By adaptively coordinating global and local observations across different task stages, FocusVLA improves multi-view utilization. On RoboTwin 2.0, FocusVLA improves the average success rate over the strongest baseline $π_{0.5}$ by 10.10 pp. In real-world evaluation, FocusVLA improves the success rate from 71.7\% to 85.0\%, achieving a 13.3\% gain over $π_{0.5}$. Moreover, the intent prediction and dynamic routing modules of FocusVLA can be integrated into existing VLA frameworks without modifying their core architectures.

0.37

Emergent Cooperative Thermoregulation in Lunar Swarm Robotics

cs.RO cs.MA ·2026-10-22 ·Anais Jeger, Harry Holt, Lennart Puck
theory
Abstract · 7 pages, 4 figures. Submitted to IEEE Robotics and Automation Letters. Video in ancillary files

Swarms of small robots are an attractive way to survey lunar caves and lava tubes. Missions such as NASA's CADRE will fly cooperating rovers for distributed mapping, but only in daylight. The 354 h lunar night drops the surface towards 95 K, and the radioisotope heaters that let single rovers survive it cannot be supplied at swarm scale. In this study we investigate whether a swarm can instead use its own members as thermal shielding. We train one policy, shared by all twenty robots, by reinforcement learning with an evolution-strategies estimator. The policy receives a single reward, once, at the end of the night: the mean temperature of the five coldest robots. Nothing in that reward refers to position, spacing, exposure or motion. Two behaviors nonetheless emerge. The swarm aggregates, and robots circulate between the sheltered core and the exposed rim, the same pair of behaviors with which emperor penguins survive the Antarctic winter. We measure the survival threshold, the smallest stored energy per robot at which every robot is still alive at sunrise. Across ten runs it is 155.6 +/- 1.3 pcm-h, where one pcm-h is 7.2 kJ of stored heat. That is 37% below a motionless huddle and within 13% of a physics-based lower bound, and at a budget where the motionless huddle loses six robots, every robot survives. The behavior needs no shared spatial reference and no robot-robot contact, and transfers from twenty robots to forty without retraining.

0.36

SEVI: Safe Evidential Real-World Reinforcement Learning

cs.RO cs.AIcs.LG ·2026-10-22 ·Shaekh Mohammad Shithil, Lantao Liu
top venuecode/data
Abstract · 13 pages, 7 figures. Accepted for publication in IEEE Robotics and Automation Letters (RA-L), 2026. Project page: https://sevi-safe.github.i…

Real-world reinforcement learning has enabled contact-rich manipulation tasks to be trained directly on physical robots, but unsafe interactions during online learning remain a key barrier to practical deployment. We propose Safe Evidential Real-World Reinforcement Learning (SEVI), a framework for reliability-aware safety intervention during real-world policy learning. SEVI introduces an Evidential Safety Critic that estimates both future constraint-violation risk and epistemic uncertainty in a single forward pass. During online learning, SEVI evaluates each task-policy action before execution; if the action is predicted to be unsafe or the prediction is insufficiently supported by evidence, SEVI overrides it with an offline-trained recovery policy that guides the robot toward a lower-risk region. We evaluate SEVI on two simulated manipulation tasks and a real-world USB insertion task under in-distribution and distribution-shifted conditions. Across task-learning and safety-critic baselines, SEVI achieves the strongest trajectory-level safety prediction, the highest final task success, and the lowest unsafe-event counts. SEVI also reduces unsafe events by 13.4 to 57.1 percent relative to recovery-based safe RL while preserving low-latency execution-time filtering. The project page is available at: https://sevi-safe.github.io/

0.36

Model Predictive Control of Wind Turbines: A Model-Oriented Review of Prediction Models, Control Objectives, and Nonlinear Aerodynamic Formulations

eess.SY math.DSmath.OC ·2026-10-22 ·Roya Khalili-Amirabadi, Mohsen Jalaeian-Farimani, Mahdi Ghane
published
Abstract · Khalili-Amirabadi, R., Jalaeian-Farimani, M., & Ghane, M. (2026). Model Predictive Control of Wind Turbines: A Model-Oriented Review of Pred…

Model predictive control (MPC) has been applied to wind-turbine regulation with markedly different internal prediction models, ranging from linearized low-order turbine models to nonlinear, aeroelastic, floating, and data-driven predictors. This review organizes the literature primarily by the model propagated by the controller and treats model order, predictor representation, MPC formulation, control objective, and validation plant as separate descriptors. Primary studies show that low-order linear or piecewise-affine predictors are attractive for constrained power and speed regulation, whereas flexible drivetrain, structural, and floating models are needed when drivetrain twist, tower motion, blade loads, or platform dynamics are explicit control quantities. Nonlinear and economic MPC studies further demonstrate coordinated pitch/torque control across broad operating regions, while preview, repetitive, and data-driven formulations address wind prediction, periodic loads, and model mismatch. Particular attention is given to aerodynamic pitch dependence: variable-pitch turbine dynamics can be non-affine with respect to applied pitch, and the term NMPC alone does not prove that the original nonlinear pitch map is retained inside the optimizer. The review therefore adopts a conservative, equation-based classification for direct nonlinear applied-pitch propagation. Here, direct nonlinear applied-pitch propagation means that the applied pitch at each prediction step is evaluated inside the original nonlinear aerodynamic map, rather than replacing that dependence by a linearized, scheduled, piecewise-affine, or learned surrogate. The literature supports a remaining integration challenge at the intersection of flexible drivetrain prediction, nonlinear aerodynamic pitch treatment, explicit constraints and load objectives, and validated adaptive or learned model correction.

0.35

Echo: Coupled Action and Task Status Signal Denoising for Vision Language Action Models

cs.RO ·2026-10-22 ·Yufeng Liu, Bohao Fan, Juntu Zhao, Xinyi Wang, Hang Yu +9 more
code/data
Abstract · Project page: https://lyfeng001.github.io/Echo/

Recognizing task completion and failure enables robots to stop upon success and initiate recovery or replanning after failure. Existing methods augment Vision Language Action (VLA) policies with task status signal prediction to support these execution decisions. However, learning and integrating these signals can require fine-grained supervision, external monitoring models, or complex architectural designs. Moreover, the potential of task status signals to directly guide action generation remains underexplored. We present Echo, a framework that combines task status signal learning without frame-level progress annotations with a compact design for status prediction and action generation within a single VLA. Specifically, Echo represents task status with stop and failure signals, added as two dimensions of each action chunk and denoised together with robot actions. Their combination distinguishes ongoing execution, successful completion, and failure. To learn these signals without fine-grained progress annotations, Echo constructs successful, failed, and recovery trajectories with complementary supervision properties. To enable status signal-guided action generation, Echo controls interactions between the signals and actions and denoises the signals faster than the actions. These allow later action denoising steps to condition on cleaner status estimates while preserving accurate signal prediction. Extensive real-world manipulation experiments show that Echo accurately predicts task status and consistently improves policy performance.

0.35

A graph decomposition approach for the synthesis of EMPs in directed acyclic networks

eess.SY ·2026-10-22 ·Eduardo Mapurunga, Michel Gevers, Alexandre Bazanella
unusual
Abstract · Paper submitted to Automatica

This paper considers the identifiability of dynamical networks whose topology is a Directed Acyclic Graph (DAG), in the context where not all nodes are excited and not all nodes are measured. The objective is to construct a valid and parsimonious Excitation and Measurement Pattern (EMP), i.e. a selection of excited and measured nodes, that allows to identify the whole network with a small number of excitations and measurements. We show first that a DAG can be decomposed into a set of trees and parallel paths networks. We propose an algorithm that achieves this decomposition and computes an EMP that combines EMPs for the individual trees and parallel paths networks. We formally show that this algorithm yields an EMP capable of identifying the whole network.

0.34

Exact Characterization of Admissible Bandwidths for Linear Disturbance Observers

eess.SY ·2026-10-22 ·Hamin Chang, Hyuntae Kim
Abstract · 5 pages, 1 figure

This paper characterizes the entire set of admissible Q-filter bandwidths, equivalently parameterized by the time constant $τ$, that guarantee robust internal stability of a linear disturbance observer (DOB) under parametric plant uncertainties. Classical small-$τ$ stability results provide a rigorous foundation for Q-filter bandwidth design. Building on this foundation, we formulate the $τ$-dependent robust internal-stability problem as a semialgebraic feasibility problem in bivariate polynomials and prove that the complete admissible set is a finite union of open intervals under the standing assumptions. A constructive procedure with exact membership tests recovers all stabilizing intervals. A numerical example with an uncertain plant family containing non-minimum-phase members illustrates a disconnected admissible set.

0.33

Input Bias Correction in Gaussian Process Regression

cs.RO cs.LG ·2026-10-22 ·Muzaffar Qureshi, Tochukwu E. Ogri, Trivikram Satharasi, Kyle Volle, Rushikesh Kamalapurkar
top venue
Abstract · Accepted for publication in IEEE Robotics and Automation Letters (RA-L) (Oct, 2026)

In traditional Gaussian process (GP) regression, the inputs are typically assumed to be known, and uncertainty enters only through the measurements. When scalar fields are mapped using GP regression with measurements obtained from a mobile sensor, the inputs correspond to the estimated pose of the mobile sensor. The goal in this letter is to understand how a recorded GP, trained on erroneous inputs (e.g., pose estimates from dead-reckoning), can be corrected in a numerically efficient way when better estimates of the inputs become available. In the developed bias-aware GP (Ba-GP) framework, the recorded GP posterior-mean gradient is used to propagate estimated input bias and noise into mean and covariance corrections. The resulting additive covariance structure is then exploited to develop numerically efficient first-order correction algorithms. A simulation study and an indoor light-intensity mapping experiment show that Ba-GP improves predictive accuracy and uncertainty calibration relative to noisy-input GP baselines, with an order-of-magnitude reduction in computational cost relative to full GP retraining.

0.33

RGraph-Eval: Task-Grounded Scene Graphs for Evaluating Robot Video Generation

cs.RO ·2026-10-22 ·Aleksandr Lazarev, Sofia Zimina, Roman Kuznetsov, Tatiana Zemskova, Iaroslav Koshelev +1 more
benchmarkcode/data
Abstract

Generated robot videos can look convincing while depicting the wrong interaction: a different object is moved, a grasp disappears, or a placement ends outside its destination. We introduce RGraph-Eval, a reference-based framework that makes these discrepancies explicit through task-grounded temporal scene graphs. Unlike approaches that construct separate evidence representations for different evaluation dimensions, we organize task entities, interactions, and motion within a unified, interpretable scene graph representation. A foundation-model pipeline identifies manipulation roles, grounds persistent object masks, infers and verifies state transitions and action intervals, and reconstructs manipulated-object motion in three dimensions. Applying this representation to a generated video and a reference demonstration supports an interpretable RGraph-Eval score with visible-object, relation-change, and 3D path-shape components. We also introduce ShareRobot-SG, a collection of 1,000 manipulation scenes from ten ShareRobot sources annotated with the multimodal representation. Our evaluation separates graph-generation quality from agreement with human video judgments and includes comparisons with adapted SVG2 and SG-Ego, as well as a benchmark of three video generators. The framework connects robot-video assessment to the observable entities and interactions underlying a task while making the graph annotations reusable beyond evaluation. Code will be publicly available at https://github.com/SashaDance/rgraph-eval.

0.33

Silence Is Not Assurance: When Production Lane Keeping Assist Does Not Warn, and What It Costs

cs.RO cs.HCeess.SY ·2026-10-22 ·Tianjia Yang, Yuxin Ding, Ke Li, Kaidi Liang, Ruwen Qin +1 more
Abstract

Production lane keeping assist (LKA) systems tell the driver whether they detect the lane markings, but whether this displayed status reflects what the system perceives, and what happens when it does not, has not been examined in real-world driving. This study introduces CANDID-LKA, a CAN-based Diagnosis of the Indicated Detection status in LKA. It evaluates a proprietary driver-facing status using three signals: the displayed status and the steering torque of the driver, both recorded from the controller area network (CAN) bus of the vehicle, and the lane perception confidence of a third-party vision system observing the same road. The framework examines how the status responds to driving conditions and estimates how completely it reports degraded lane perception by comparing it with the other two signals in a latent-class model. We apply it to 39,800 seconds of real-world driving from 13 vehicle models of three manufacturers. The results show that the status does not respond consistently to the conditions that make markings harder to see, and that the three makes differ widely in how often they report not detected. The latent-class model shows that the displayed status is incomplete: it reports only 47% of degraded seconds as not detected, compared with 80% for the third-party low perception confidence, and in half of the low-confidence episodes it never reports not detected. In the silent seconds in which the status reads detected while the perception confidence is low and the driver is not in control, the standard deviation of lateral position over the next 3 s is about 60% larger than when the status and the perception confidence agree. A detected status thus does not guarantee reliable lane perception, and silence with the driver not in control is the case that most needs attention.

0.31

Distributed Swarm Morphogenesis with Neural Cellular Automata

cs.RO cs.MA ·2026-10-22 ·Ray Verma, Anthony Tzes, Eliseo Ferrante
Abstract

Swarm morphogenesis requires robots to assemble desired shapes through local interactions. We present a modular framework integrating neural cellular automata (NCA) with existing site assignment and collision-aware control. Within this framework, we introduce a decentralized attachment-site discovery mechanism. Robots already in the formation update and share local NCA states to identify new attachment sites, enabling progressive growth without a predefined list of sites during deployment. To bridge the mismatch between conventional NCA growth and gradual physical assembly, we additionally propose offline sparse-retention fine-tuning that simulates partial attachment, together with a false-positive loss that penalizes invalid site activations. Simulations demonstrate formation of diverse lattice morphologies, scaling to several hundred robots, and substantial improvements in morphology accuracy from the proposed fine-tuning. Experiments with Crazyflie quadrotors validate the integrated execution pipeline using robot-relative observations reconstructed from external tracking.

0.31

Half-Slot Link Scheduling for Open GNSS with $x$-Tx $y$-Rx Multibeam Phased Array Antennas

eess.SY ·2026-10-22 ·Huan Yan, Kanglian Zhao, Wenfeng Li, Jinjun Zheng, Ziyu Wang +1 more
Abstract · 20 pages and 10 figs

Global Navigation Satellite Systems (GNSSs) are evolving from closed navigation constellations toward open space infrastructure capable of serving extensive spacecraft users. This evolution requires the constellation to support navigation ranging, telemetry (TM) return, and telecommand (TC) delivery for users while preserving its own ranging and communication functions.Existing phased-array payloads onboard BeiDou satellites provide one transmit (Tx) beam and one receive (Rx) beam. Such a configuration was designed mainly for intra-constellation operation, and can become a resource bottleneck when a large number of users share the same link opportunities. Upgrading GNSS satellites to phased-array terminals with $x$-Tx and $y$-Rx beams is therefore a natural enhancement. This paper develops a staged dynamic half-slot link scheduling framework (SDHSS) for open GNSS with $x$-Tx $y$-Rx multibeam phased-array antennas. The framework models navigation, TM, and TC as three different services, reserves the essential ranging and ground-access opportunities required by the GNSS constellation, and then allocates the remaining half-slot beam resources according to current service demand. To support multibeam concurrency, feasible links are selected through dynamic service-layer ordering and beam-copy matching. Simulations based on a BeiDou constellation show that the proposed method protects GNSS baseline operation and provides stronger external navigation, TM, and TC performance than fair contact planning. The results further show that additional Tx and Rx beams can be effectively converted into multi-service capability. This work establishes a link-scheduling approach for extending GNSS from constellation-internal operation to large-scale spacecraft service support.

0.30

UniFlowWAM: A Unified World-Action Model with Visual Semantics and 4D Geometry

cs.RO ·2026-10-22 ·Jialin Tian, Zhuofan Li, Xinkai Wang, Bokai Lin, Chenyi Wang +3 more
code/data
Abstract · 14 pages, 5 figures, 4 tables. Project page: https://kksham111.github.io/UniFlowWAM_ProjectPage/

Robot demonstrations record what the robot saw and what it did. They do not specify how the 3D world changed in between. Scene-level 3D flow describes it by tracking 3D scene points, providing the world representation with 4D geometry. Prior work has used flow to improve robot policies. Flow served as an intermediate prediction that a separate model converts into actions, as a prior injected at inference time, or as an extra output learned together with the action. In all of these designs, flow stays outside the policy, as a separate module or as an extra prediction head. We introduce UniFlowWAM, a Unified World-Action Model that makes 3D scene flow a third modality on par with vision and action. Its world representation therefore covers both visual semantics, through future visual features, and 4D geometry, through 3D flow, and flow receives its own supervision. UniFlowWAM is a diffusion transformer with one expert per modality, and the three experts exchange information through joint attention. We train UniFlowWAM under a ten-objective role grid, where each objective assigns every modality exactly one of three roles, namely target, condition, or absent. The same model thus serves as a policy, as forward and inverse dynamics, and as a predictor of future visual features or 3D flow. We evaluate UniFlowWAM on RoboCasa-GR1, LIBERO, and LIBERO-Plus in simulation, and on four real-world bimanual manipulation tasks. On RoboCasa-GR1, UniFlowWAM raises the success rate of its base model without flow from 52.1% to 56.4%. It preserves in-distribution success on LIBERO at 98.1% and reaches 87.7% zero-shot on LIBERO-Plus. On the real robot, it achieves 87.0% average success, compared with 68.5% for the strongest baseline.

0.30

Learning-Oriented Fault-Tolerant Control for UAVs with Time-Varying Fault Estimation

eess.SY ·2026-10-22 ·Moh Kamalul Wafi
published
Abstract · This is an extended version of the paper published in the proceedings of IEEE CoDIT

The increasing deployment of unmanned aerial vehicles (UAVs) in safety-critical applications has created a growing demand for control strategies that remain reliable in the presence of actuator and sensor faults. This paper presents a fault-tolerant control framework that combines constructive mathematical modeling, proportional--integral (PI) fault estimation, and learning-oriented control. A nonlinear quadcopter model is first developed using both the Newton--Euler and Euler--Lagrange formulations, from which a discrete-time linear model is derived for observer and controller design. A PI observer is then proposed to estimate actuator and sensor faults, with exponential estimation-error convergence established for constant faults in the absence of disturbances and ultimate boundedness established for slowly time-varying faults under bounded disturbances. The estimated fault signals are incorporated into an active fault-tolerant control architecture together with a Deep Deterministic Policy Gradient (DDPG) controller, which is trained using only the yaw channel and subsequently applied to improve closed-loop tracking performance. Numerical simulations illustrate the observer's estimation performance under constant and slowly time-varying faults, including bounded disturbances, and the tracking performance of the integrated controller under multiple actuator and sensor fault scenarios. The proposed framework combines model-based fault estimation with learning-based control to improve fault accommodation and tracking performance in UAV systems.

0.29

Tilt-Rotor Quadcopter with Learned Residual Control and Mode-Dependent State Estimation

cs.RO ·2026-10-22 ·Ninad Joshi, Prabhat Kumar
Abstract · 5 figures

Tilt-rotor quadcopters redirect thrust for translation without banking, but complicate modeling, state estimation, and disturbance rejection. This paper presents a simulation-based flight stack for a \SI{2.00}{\kilo\gram} tilt-rotor quadcopter. The system uses a six-degree-of-freedom nonlinear model with first-order motor dynamics and two-axis rotor tilt, parameterized using a 3D CAD model. A cascaded PD controller commands forward and lateral tilt angles. A 14-state EKF fuses sensor data across hover, early-transition, and transition modes using a hysteretic (Schmitt-trigger) classifier to prevent covariance discontinuities. Because the proportional position loop leaves a steady-state offset under constant wind, a small bounded neural residual is added on top of the unmodified PD controller. It is trained by behavioral cloning from a teacher with privileged access to the true wind, while the deployed network never observes wind. In simulation, the EKF achieves a post-settling position root mean square error (RMSE) of \SI{0.226}{\metre} (horizontal) and \SI{0.03}{\metre} (vertical) and an attitude RMSE below \SI{0.8}{\degree}, and it estimates a wind velocity of $(\num{2.0}, \num{-1.2})\,\unit{\metre\per\second}$ to within \SI{0.15}{\metre\per\second} per axis. The residual reduces steady-state station-keeping error under wind by approximately \SI{49}{\percent}, from \SI{0.117}{\metre} to \SI{0.060}{\metre} at \SI{1.5}{\metre\per\second} and from \SI{0.319}{\metre} to \SI{0.162}{\metre} at an out-of-distribution \SI{3.9}{\metre\per\second}. The framework provides a basis for hardware-in-the-loop and flight-test studies.

0.29

TEAR: Task-Evidence Authority Routing for Robotic Manipulation

cs.RO ·2026-10-22 ·Shicheng Zhang, Hui Liu, Haoliang Li
Abstract · 9 pages

Reliable robotic manipulation in field-deployed settings requires more than executing a capable controller. As the execution conditions change, the system must decide, online and on the robot side, which control source should hold authority. Human teleoperation adapts readily to contact-rich and semantically ambiguous situations but depends on a stable communication link. Local autonomy is executed without continuous human intervention, but may fail under semantic ambiguity, long-horizon drift, and unforeseen contingencies. These two control sources exhibit complementary reliability profiles. We formulate this complementarity as a control authority routing problem and present Task-Evidence Authority Routing for Robotic Manipulation (TEAR), an edge-deployable framework that allocates authority between human teleoperation and local autonomy. TEAR arbitrates on the robot side using network-state evidence, which governs the usability of teleoperation, and task-semantic evidence, which reflects whether autonomous execution remains consistent with the current task. A lightweight temporal arbitrator observes both forms of evidence together with robot state and current authority to time authority transfers. Across simulated and real-world trials under dynamic network conditions, TEAR achieves task success rates close to pure teleoperation while limiting teleoperation to 19.8-27.4% of control steps and improves success by 30-40% over autonomy-only execution in our settings. These results demonstrate the potential of complementary authority routing to sustain successful manipulation with substantially reduced dependence on continuous human control.

0.29

Anomaly Score Preserving Data Poisoning Attacks on Data-Driven Controller Design

eess.SY ·2026-10-22 ·Motoharu Tsuboi, Hampei Sasahara, Iori Takaki, Hideaki Ishii
Abstract · 8 pages, 3 figures, 2 tables

This paper investigates stealthy data poisoning attacks against data-driven controller synthesis and develops an attack generation method. During data collection, an attacker can sequentially manipulate state measurements, while the defender attempts to detect such manipulation using a residual-based anomaly score based on the Mahalanobis distance. Under suitable assumptions on the defender's prior knowledge, we characterize the class of time-varying affine transformations that preserve the anomaly score for arbitrary data and parameterize this class by the orthogonal group. Based on this characterization, we formulate attack design under the stealthiness condition as an optimization problem over the orthogonal group and propose a Riemannian gradient method. Numerical experiments demonstrate that the proposed attack can degrade closed-loop performance while remaining stealthy.

0.28

Model-Free Generative Stealthy Attacks on Safety-Critical Vehicle Cyber-Physical Systems

eess.SY ·2026-10-22 ·Qazi Mairaj ud din, Sidra Ghayour Bhatti, Qadeer Ahmed
Abstract

Existing stealthy attacks presuppose white- or grey-box access to the target system, unrealistic in practice. This paper develops a model-free stealthy attack framework that learns an attack policy directly from nominal measurement data, without knowledge of the system dynamics, estimator, or detector. We establish a certified residual-impact bound and an observable stealth budget around a stealth-consistent anchor, and show that recurrence and context conditioning are structural necessities for closed-loop, multivariate cyber-physical systems (CPS). The resulting attack generator, trained against a Wasserstein critic under a stealth-constrained impact objective, induces a hazardous 2.51 m lane departure while evading five intrusion detection systems, and, in real-time hardware-in-the-loop tests, a speed deviation of approximately 8 km/h while staying below the detection threshold. Results show that safety-critical CPS remain exposed without attacker access to privileged information, motivating offense-informed defenses.

0.28

FastHarness: Experience-Driven Review Scheduling for Vision-Language-Action Policies

cs.RO ·2026-10-22 ·Tao Lin, Yuxin Du, Haoxiao Wang, Zile Zhong, Bo Zhao
code/data
Abstract · Github:https://github.com/FastHarness/fast-harness

Agentic robot systems combine language-model reasoning with robot policies to monitor execution and recover from failures. However, reviewing every action chunk incurs substantial overhead, even in familiar situations. We introduce FastHarness, an experience-driven execution harness that adaptively schedules GPT-6 Astra supervision of a π0.5 policy. A fast path uses stage memory and reviewed outcomes to reduce review frequency as experience accumulates, while a complementary recovery path focuses supervision on correcting errors. Once recovery is verified, routine execution resumes, with failure experience retained to guide future interventions. The policy remains frozen and generates fresh actions from current observations. Across four simulated manipulation tasks, reviewer token use decreases by an average of 57.2% per task from the first evaluated episode to the mean of the final five episodes. Mean episode time decreases from 85.5 to 52.0 minutes over the same interval. On the same tasks, mean success increases from 80% for GPT-as-Policy (π0.5 + GPT-6 Astra) to 97.19% with FastHarness. These results highlight adaptive supervision as a practical route to more efficient agentic robotics without additional model training.

0.28

A Sparse Topometric Graph for Scalable Cross-Session Indoor Navigation

cs.RO cs.CV ·2026-10-22 ·Eun Sun Lee, Minkyun Seo, Gwangtak Bae, Junho Kim, Seunggu Kang +2 more
hot direction
Abstract

A conventional embodied agent in an indoor environment builds a map by accumulating noisy real-world measurements and results from heavy inference to guide navigation. While the detailed measurements may differ due to scene changes or sensor settings, the unique topological structure of navigable passages or the connectivity between rooms may provide training-free guidance that complements local observations. We propose a sparse topometric graph as an efficient structural representation for robust indoor navigation, alleviating the need for fine-grained matching, which is computationally expensive, memory-intensive, and prone to real-world variations. The map consists of nodes that encode local 3D geometry and distance-weighted edges that capture traversable connectivity. A learned graph-matching network finds consistent subgraph structures across sessions by considering both node-level geometric descriptors and neighborhood connectivity. The holistic contextual matching is robust to local ambiguities caused by weak distinctiveness or repetitive structure, and establishes stable correspondences despite local scene changes, different sensor modalities, or limited spatial overlap. We analyze the practicality of the graph through cross-session point-goal navigation, where a goal specified in a prior reference map must be reached from an unknown initial pose in a later session. Experiments show that our method achieves the highest success rates (77.1% on Gibson and 40.0% under 3D-FRONT scene changes), while reducing memory by more than two orders of magnitude compared to point-level representations. The proposed approach further supports cross-session alignment and goal-directed navigation with strong scalability and robust real-world transfer.

0.28

KiRoSe: Using Kinematic Data for Efficient Segmentation of Robot Demonstrations

cs.RO ·2026-10-22 ·Shuangyu Xie, Kaiyuan Chen, Muhammad Zubair Irshad, Richard Cheng, Sergey Zakharov +1 more
Abstract

Segmenting robot demonstration data into disjoint temporal and semantic subsegments is valuable for learning, cleaning, retrieval, and analysis. Segmentation is often performed manually in a tedious and error prone process. Multimodal Large Language Models (MLLMs) have the potential for automating this process, but directly ingesting raw multi camera video streams into MLLMs can be costly and can quickly exhaust context windows. In this paper, we introduce KiRoSe (Kinematics for Robot Segmentation), a novel framework that focuses on the companion stream of robot kinematic data. Inspired by the principle of classical entropy coding, such as Huffman coding, where frequently occurring symbols are represented more compactly, we propose Sparse Motion Tokens (SMTs), a discrete kinematic vocabulary designed to compactly represent recurring, temporally localized motion patterns, which we refer to as kinematic events. SMTs are learned by encoding kinematic trajectories with temporal segment supervision and quantizing the resulting features. We use increasingly available robot demonstration subtask annotations to supervise an open language model to interpret the resulting token sequences. Evaluation against human labeled segmentation shows that KiRoSe can substantially outperform baseline segmentation methods in F1 quality and transfers zero shot to unseen robot embodiments and egocentric human motion, while preserving the language models open vocabulary knowledge. Beyond segmentation, KiRoSe tracks task progress on an unseen robot and detects when an action is being reversed.(website: https://berkeleyautomation.github.io/kirose/)

0.27

In-Context Learning World Models for Robot Dual-Arm Manipulation of Soft Garment with Partial Observations

cs.RO ·2026-10-22 ·Hengyi Sim, Quentin Rouxel, Dennis Hong, Darwin Caldwell, Fei Chen
Abstract

Robust manipulation of soft objects, such as garments, is a critical skill for service and household robots to adopt, but remains challenging due to its high-dimensional state space, and material-dependent deformations. Humans address this challenge by using mental world models that "imagine" the shape and material properties from partial observations, which enables them to predict how actions affect future states. As such, embedding such world models in robots allows for better generalization in manipulating a wide variety of garments in the real world. However, learned world models typically require large amounts of real world data that is strenuous to collect, and a mechanism to infer latent properties that govern dynamics. We present a Soft-Object World Model that performs generative full-state estimation and action-conditioned dynamics prediction using a transformer backbone trained on large-scale simulated soft object interactions, with measures to mitigate the sim-to-real gap. By learning to infer latent shape and material cues through in-context learning over a variable-length partial point-cloud history, the model generalizes across robot embodiments, material parameters, garment types and shapes, and manipulation coordination modes without weight updates. The world model is then integrated with a Goal-Conditioned action planner for motion synthesis. We validate the approach in simulation and real world for bimanual garment folding, reporting improvements in reconstruction fidelity, prediction accuracy, and task success. Project Website: https://1am5hy.github.io/ICL soft object world model/

0.27

On the Structure of Quantum Finite-Level Systems: Decoherence-Free Subsystems, Quantum Nondemolition Variables, and Back-Action-Evading Measurements

eess.SY quant-ph ·2026-10-22 ·Jia Gu, Guofeng Zhang, Ian R. Petersen
Abstract · 16 pages, comments are welcome

The purpose of this paper is to conduct a comprehensive structural analysis of quantum finite-level systems in the Heisenberg picture, where the evolution of system observables is modeled by bilinear quantum stochastic differential equations (QSDEs). Firstly, real coordinate transformations that preserve commutation and anticommutation relations are characterized, and relative admissibility between different orthonormal Hermitian bases is formulated. Secondly, strongly aligned subspaces are introduced, based on which decoherence-free subsystems (DFSs) are constructed in general orthonormal Hermitian basis as well as the generalized Gell-Mann basis. Thirdly, conditions for the existence of quantum nondemolition (QND) variables are given in terms of both linear systems theory and spectral separation by output-signal observables. Finally, a sufficient condition for quantum back-action-evading (BAE) measurements is established. A four-qubit nearest-neighbor Heisenberg chain with collective dephasing illustrates these structural properties of quantum finite-level systems.

0.27

A Transparent 3-DoF Assistive Device for Upper-Limb Tremor and Ataxia Compensation: Design and Experimental Validation

cs.RO ·2026-10-22 ·Thomas Dobosz, Veronika Hofmann, Ahmed Barhoma, Ann-Kathrin Henke, Johannes A. Gaus +10 more
Abstract · 8 pages, 7 figures, 3 tables, Preprint

Assistive devices for patients with neurodegenerative movement disorders must combine high mechanical transparency with the ability to generate corrective interaction forces. Achieving both properties simultaneously remains challenging. This paper presents the completed iAssistADL platform, a 3-DoF assistive device for future tremor and ataxia compensation during activities of daily living. The system employs a transparency-oriented mechanical architecture with all actuators located at the base, resulting in low moving mass, decoupled kinematics, and high mechanical transparency. To compensate for friction-induced transparency limitations along the vertical axis, a hybrid control strategy is proposed that combines open-loop force rendering in the horizontal plane with closed-loop force control in the vertical direction. The platform is experimentally evaluated through transparency measurements and static and dynamic force-rendering experiments at representative configurations. Results demonstrate resistive torques below 0.1 Nm in the horizontal plane, end-effector forces of up to approximately 10 N throughout the workspace, and accurate horizontal force rendering up to 4 Hz. While the vertical axis remains limited by friction-induced dynamics at higher frequencies, the proposed system successfully combines high transparency with controlled force generation, establishing a promising hardware platform for future tremor and ataxia compensation.

0.27

Space Solar Applications: Simulation with MATLAB-Simulink Environment and Transmission of Solar Power Generated

eess.SY ·2026-10-22 ·Ferhat S. Demirci, Murari Koushik Muthu Kannapiran
surveyunusual
Abstract · 6 pages, 16 figures, 4 tables. Work done at Illinois Institute of Technology, 2010

In this paper a photovoltaic (PV) array suitable for use in outer space is simulated in the MATLAB/Simulink environment. The simulation of the PV array is first carried out with maximum power point tracking (MPPT) and then without MPPT. In a second step, the response of the system to different irradiation levels is observed. A brief overview of the transmission of solar power generated in space down to the earth via microwaves is also given. Selected simulation results are provided for a comparative study of both cases.

0.26

Robot Motion Transition with a Human Motion Prior

cs.RO ·2026-10-22 ·Taehyun Byun, Taemoon Jeong, Yoonbyung Chai, Woojin Wie, Insu Park +2 more
code/data
Abstract · 9 pages, 4 figures

Humanoid motion-tracking policies can reproduce diverse behaviors, yet connecting independently acquired motions into a continuous sequence remains challenging. Interpolation connects boundary poses without considering whether their relative position, heading, and timing provide suitable conditions for a transition. A generative motion prior can produce richer motion, but poorly arranged boundaries may cause it to generate excessive locomotion, turning, or another unintended action. We present Spatio-Temporal Alignment with a Motion Prior (STAMP), a framework for connecting independently acquired humanoid motions through a Robot-to-Human-to-Robot process. STAMP uses human motion space to draw on rich motion priors and reuse transition knowledge across robot embodiments. Its key component is a learned alignment model that determines the boundary conditions for motion inbetweening. Given the two motions retargeted to a human rig, the alignment model ranks candidate positions, headings, and transition durations for the following motion. A human motion prior generates the transition under the selected alignment. The completed sequence is retargeted to the robot and used as a reference to train a motion-tracking policy. We evaluate the tracking success of the generated transitions on Unitree~G1 and~ROBOTIS K1. Physical robot demonstrations further show that STAMP connects independently acquired motions into long-horizon sequences. Project page: https://childtoy.github.io/STAMP-project

0.26

Can An Array of Magnetometers Replace a DVL - A Cramér-Rao Bound Analysis

eess.SP ·2026-10-22 ·Ludvig Lindström, Isaac Skog, Magnus Lundberg Nordenvaad, Andreas Gällström
Abstract · Submitted to IEEE Journal of Ocean Engineering, 9 pages, 5 figures

Underwater vehicles rely on noisy dead-reckoning systems, such as Doppler velocity log (DVL)-aided inertial navigation systems, for positioning. However, DVLs rely on active acoustic measurements and are thus unsuitable for covert operations. Magnetic velocity estimation using a magnetic sensor array is a potential silent alternative. In outdoor environments, spatial variations in magnetic field magnitude are small, whereas temporal variations are large, making the achievable velocity-estimation accuracy uncertain. In this paper, Cramér-Rao lower bounds (CRLBs) are derived for magnetic velocity estimation using different sensor configurations and measurement models in one dimension, ignoring platform disturbances, to quantify the potential accuracy of magnetic sensor arrays. The analysis suggests that, when spatial variations are sufficient and temporal variations can be corrected using correction data, the resulting CRLB is comparable to that of an export-controlled DVL. If, instead, temporal variations are not corrected, the sensor noise must be reduced by around two orders of magnitude for the CRLB to reach the accuracy of an export-controlled DVL. Therefore, the results indicate that current magnetic sensor arrays may complement, rather than replace, DVLs in inertial navigation systems.

0.26

AstroBench: Benchmarking Free-Floating Capture and Space Robotic Manipulation

cs.RO ·2026-10-22 ·Zizhe Wang, Jia Deng, Xin Lu, Zehao Jia, Yi Huang +1 more
benchmarkcode/data
Abstract · 12 pages, 5 figures, 9 tables. Code and assets: https://github.com/AstroUpLab/AstroBench ; project website: https://astroup.com/astrobench/

Capturing a loose object inside a spacecraft requires a robot to approach a target that can translate and rotate, and whose motion changes when contact is made. These conditions motivate evaluation beyond manipulation of supported objects. We introduce AstroBench, an open benchmark for free-floating capture and space robotic manipulation with a fixed-base arm. The task suite combines four initial-motion tracks with sample-storage operations based on MELFI hardware and cap and captive-screw servicing based on the Robotic Refueling Mission. The two workcells provide nine task definitions and 29 task-target combinations. A metadata-filtered Objaverse inventory supplies 8,414 object instances for asset expansion, alongside five cabin visual contexts and versioned task instructions. We report two acquisition-focused experiments using existing checkpoints. A frozen $π_{0.5}$ DROID joint-position policy captures 2 of 200 static targets and 4 of 200 moving targets across twenty object identities. A cube-specific ACT policy captures 8 of 10 static targets and 6 of 10 moving targets in a separate experiment. Each experiment retains its recorded observation, action and capture configuration. The results document difficult direct transfer for the tested generalist checkpoint and successful task-specific capture behavior, while identifying concrete settings for studying motion-aware perception and contact-sensitive control. Code, task definitions and editable workcell assets are available at https://github.com/AstroUpLab/AstroBench.

0.26

Progress Probes $P^2$: Lightweight Runtime Monitoring for Vision-Language-Action Models

cs.RO ·2026-10-22 ·Atiksh Bhardwaj, Edward W. Duan, Junwon Seo, Prithwish Dan, Wei-Chiu Ma +2 more
code/data
Abstract

Vision-Language-Action (VLA) models can perform diverse manipulation tasks directly from visual observations and natural language instructions, but they remain prone to mistakes and unpredictable failures. Deploying these policies without continuous human supervision will require robust runtime monitors that detect when the policy has gone off-course. Existing monitors use external foundation models, such as vision-language models (VLMs) to supervise the policy, introducing significant complexity and computational cost for monitoring at runtime. In this work, we investigate an alternate approach: \textit{can a VLA's own internal activations provide the information needed to monitor its execution?} Our method trains linear probes on policy activations from demonstrations to estimate progress, then flags executions that deviate from expected progress. In experiments with the open-source VLA $π_{0.5}$ and across simulated and real-world manipulation tasks, this approach is competitive with or outperforms VLM-based monitors in task failure detection, as well as methods which require failure data for training. These findings suggest that simple probes of VLA internals can be used to develop reliable runtime monitors, opening a path toward building policies that can act reliably and monitor their execution without external foundation models.

0.26

Energy-Efficient Risk-Constrained Adaptive Semantic Communications Under Uncertain Adversarial Threats

eess.SP ·2026-10-22 ·Maedeh Fallahreyhani, Paeiz Azmi, Nader Mokari, M. Reza Abedi
Abstract

Semantic communication systems based on deep joint source-channel coding can provide task-oriented wireless transmission, but adversarial robustness may require substantial radio and receiver-side defensive-computation energy. This paper develops an energy-efficient, risk-constrained semantic communication framework under source-domain, over-the-air (OTA), and potentially composite adversarial conditions. Long-term total energy, including communication and diffusion-purification energy, is minimized subject to average semantic-loss, trajectory-level tail-risk, and end-to-end latency constraints. A recurrent primal-dual proximal policy optimization controller jointly adapts transmit power, semantic/channel bandwidth ratio, and purification complexity using causal channel observations, semantic confidence, imperfect attack-belief estimates, and recurrent history. A conditional value-at-risk (CVaR) constraint explicitly limits severe semantic failures. On CIFAR-100, at the unseen source-domain level $ε=12/255$, the proposed controller achieves approximately $61\%$ robust Top-1 accuracy with $\mathrm{CVaR}_{0.9}\approx0.36$ while using about $9\%$ less total energy than diffusion-only defense. At the unseen OTA level $Γ_A=-5$ dB, it achieves approximately $52\%$ robust Top-1 accuracy, reduces semantic tail risk by $40\%$, and consumes about $8\%$ less total energy than diffusion-only defense. These results demonstrate energy-efficient semantic defense under partially observable adversarial conditions.

0.24

Constrained Sparsest Input Selection for Structural Controllability is NP-Hard

eess.SY ·2026-10-22 ·Yuan Zhang, Jintao Zhang, Yun Feng
Abstract · 6 pages, 1 figure

We study the problem of selecting the fewest admissible input--state links that make a structured linear system structurally controllable. This objective counts nonzero entries from prescribed positions of the input matrix and is different from minimizing its number of nonzero columns. Prior work gives polynomial-time algorithms for several restricted cases, including single or dedicated inputs and prescribed input patterns satisfying additional graph-theoretic conditions. We give a polynomial-time reduction from perfect three-dimensional matching, establishing for the first time that the decision problem is NP-complete and the optimization problem is NP-hard. A refinement of the reduction yields an exact optimum relation and shows that the problem is APX-hard. An example also exhibits a strict integrality gap for the natural linear programming relaxation.

0.24

Probabilistic Energy Modeling Unifies Versatile Swarm Control

cs.RO cs.MA ·2026-10-22 ·Xiang Zhou, Xiangyu Zhu, Xintong Wang, Qingyuan Yang, Rui Huang +3 more
Abstract · 25

Robotic swarms must coordinate across diverse missions, yet existing controllers often require redesign, retuning, or retraining when objectives change. Here we show that probabilistic energy modeling provides a unifying representation for versatile swarm control, in which diverse missions emerge from compositions of shared interaction principles rather than task-specific control formulations. By exploiting the correspondence between Gibbs random field energy minimization and structured reward maximization, we encode fundamental swarm behaviors as reusable energy terms within a unified probabilistic objective. Decentralized policies trained within this representation are composed into a single controller through lightweight policy gating, without retraining. An interpretable action-attention mechanism captures non-uniform local interactions, while dimensionless normalization supports transfer across swarm scales and platforms. Demonstrated on aerial swarms, the unified policy generalizes zero-shot across six mission classes, 10-100 agents, two hardware platforms, and indoor cluttered and GPS-denied woodland environments. Large-scale simulations achieve 0.96-1.00 task success rates with 1-10 ms per-agent inference latency, while physical deployments confirm zero-shot transfer to real platforms and unstructured environments. These results establish probabilistic energy modeling as a scalable and transferable framework for decentralized swarm control.

0.24

Asset-Normalized Distribution Reliability Budget Optimization Using Historical Outage Benefits and Work-Type Learning

eess.SY ·2026-10-22 ·Hritik Gopal Shah, Catherine Tajmajer
Abstract

Reliability investments compete for constrained capital but are often evaluated in separate programs. This paper presents a data-driven framework that translates historical work orders, geographic asset records, outage events, and costs into comparable customer-minute and customer-interruption benefits at the circuit-zone level. A normalization process maps historical device names and non-isolating assets to current automaticisolating-device zones and links investments to pre- and postconstruction outage histories. Utility data covering more than $250 million in investments completed from 2010 to 2020 are used to estimate benefit factors for seven work types. Those factors feed a mixed-integer portfolio model that selects at most one applicable action per zone under a capital budget. Observed reductions range from 32% to 94% for customer-minutes interrupted and from 43% to 93% for customer interruptions. The framework provides a traceable path from heterogeneous utility records to zone-level reliability investment decisions

0.24

Numerical Solution of Large-Scale Low-Rank Coupled Generalized Sylvester Equations

eess.SY math.NA ·2026-10-22 ·Umair Zulfiqar
benchmark
Abstract

This paper develops projection methods for computing low-rank solutions to large-scale coupled generalized Sylvester equations. The dominant subspaces of these solutions are shown to be spanned by specific resolvent directions, motivating a Petrov--Galerkin projection framework. Conditions for the unique solvability of the projected coupled generalized Sylvester equations are also discussed. Based on these dominant subspaces, two algorithms are proposed: an extended Krylov-based projection algorithm and a rational Krylov-based projection algorithm. The former uses fixed shifts, whereas the latter uses adaptive shifts. Numerical experiments on large-scale benchmark problems demonstrate that the proposed methods achieve high accuracy and fast convergence.

0.23

Consistent Distributed Invariant EKF SLAM with Point and Object Features

cs.RO ·2026-10-22 ·Mengya Xu, Xianjia Yu, Ha Sier, Shoudong Huang, Tomi Westerlund
Abstract · Submitted to IEEE Robotics and Automation Letters (RA-L)

Multi-robot SLAM allows robots to improve their estimates by exchanging map information, but repeated exchanges create unknown correlations between local estimators. Covariance intersection (CI) can fuse estimates without tracking these correlations. In right-invariant extended Kalman filter (RI-EKF) SLAM, however, point features obtain their invariant errors through the robot rotation, so a neighbor's map estimate is expressed in its own error coordinates and is correlated with its pose, and cannot be fused by CI directly. This paper presents Joint-CI, a unified fusion method for distributed RI-EKF SLAM with point, object, and mixed point--object maps. Using a neighboring robot's local posterior and an inter-robot observation, a joint estimate of the receiving robot's pose and the common features is constructed in the receiving robot's error coordinates, retaining the robot--feature and feature--feature correlations. Partial-state CI then fuses this estimate with the receiving robot's local posterior. We show that the unified formulation and the fusion preserve the six unobservable directions associated with global rotation and translation, and that communication and fusion preserve first-order covariance consistency given consistent local posteriors. Monte Carlo simulations and a real-world dataset-derived experiment show that Joint-CI improves estimation accuracy while maintaining consistent, mildly conservative uncertainty.

0.23

Large Language Models Meet Bayesian Optimization: Towards Self-evolving Wireless Digital Twins

eess.SY cs.IT ·2026-10-22 ·Yingzhuo Sun, Haoyun Li, Ming Xiao, Francisco J. Ferrao, Berna Ozbek
Abstract

Wireless digital twins (DTs) are increasingly integrated with artificial intelligence to support online modeling, prediction, and network optimization. However, most existing approaches perform adaptation within predefined learning or optimization structures, which may become unsuitable when wireless tasks and operating conditions change. In this article, we investigate the integration of large language models (LLMs) and Bayesian optimization (BO) to extend wireless DTs from online adaptation towards algorithm-level self-evolution. We introduce an LLM-assisted BO framework in which BO provides a stable and sample-efficient optimization backbone, while the LLM determines whether selected BO components should be retained or adapted based on the current task and accumulated optimization experience. In simulations, we apply the framework to Gaussian process (GP) kernel discovery for a multi-user wireless beam-selection problem. Simulation results show that the LLM-discovered kernel improves the final BO performance by 19.7% on the training tasks compared with the strongest standard kernel baseline. The results demonstrate the potential of combining LLM flexibility with structured BO for self-evolving wireless DTs.

0.23

Electromagnetic Modeling of Deformation-Based Pinching Antennas

eess.SP ·2026-10-22 ·Dimitrios Tyrovolas, Nikos C. Sagias, Ke Guan, George K. Karagiannidis, Stylianos D. Asimonis
Abstract

Pinching-antenna systems (PAS) rely on local perturbations of a guiding structure to radiate part of the guided electromagnetic energy, making the physical response of the resulting pinching antenna (PA) fundamental to both guided-wave propagation and free-space radiation. To capture this coupled behavior, we develop an electromagnetic model of a finite deformation-based PA that directly relates the applied waveguide perturbation to its complex reflection and transmission responses and three-dimensional radiation characteristics. Full-wave MEEP simulations validate the model and demonstrate how the pinching ratio and pinched-section length jointly affect the transmitted power and phase, radiation power-fraction, and radiation pattern, thereby establishing PA geometry as a degree of freedom for controlling the radiation response. Importantly, this geometric dependence extends to communication performance, as the received signal-to-noise ratio (SNR) results reveal that neither minimizing the PA--user distance nor maximizing the radiation power-fraction necessarily yields the highest SNR, since the applied perturbation also determines the realized gain toward the user.

0.23

GelSlim 5.0: Unifying Vision and Touch with a Single Fingertip Camera

cs.RO ·2026-10-22 ·Xiangdong Wang, I-Hsuan Huang, Xiaohao Xu, Xili Yi, Nima Fazeli +1 more
Abstract

Vision and touch provide complementary cues for object pose and local contact geometry, but external cameras are often occluded at the contact site. We present GelSlim 5.0, a fingertip sensor that acquires visual and tactile observations with one fixed camera and a continuous UV-responsive contact layer. By switching illumination, the camera sequentially acquires through-membrane visual images and contact-sensitive tactile images from a shared viewpoint, avoiding inter-camera extrinsic calibration. To combine these observations, we introduce VisTacFusion, a multimodal transformer with separate prediction paths for planar object pose and contact geometry. Asymmetric bottleneck fusion incorporates visual context while preserving tactile spatial features for depth and surface-normal estimation. Modality dropout enables inference with paired inputs or either modality alone. On real validation data from 12 known objects, paired inputs reduce depth MSE, angular error, and translation $L_1$ by 61.2%, 72.3%, and 61.3%, respectively, relative to tactile-only inference with the same model. Relative to visual-only inference, the corresponding reductions are 93.7%, 6.1%, and 8.6%.

0.23

Practical Complete Tracking Through Gimbaled Unidirectional Actuation with Application to Multirotor UAVs

eess.SY ·2026-10-22 ·Ian Willebeek-LeMair, Craig Woolsey
Abstract · 16 pages, 10 figures

This paper addresses the problem of practical tracking control for robotic vehicles with gimbaled unidirectional actuators, with special application to multirotor unmanned aerial vehicles. A widely recognized obstacle to tracking control for multirotors is the free-fall singularity --- when a multirotor is commanded to free fall, the vehicle's target attitude becomes undefined. This singularity is a pathology pertinent to all robotic vehicles with gimbaled unidirectional actuators. This paper considers the general theoretical problem for vehicles with gimbaled actuators. Starting from a baseline robust controller that assumes unconstrained inputs, we redesign the control law to be compatible with the gimbaled actuator. This is accomplished by driving the output of the gimbaled actuator to a ball centered at the target input value. The baseline controller's tracking guarantees are recovered in a practical sense. We apply our results to synthesize a global practical complete position tracking controller for multirotor unmanned aerial vehicles. The theory is demonstrated in simulation.

0.23

Multimodal MPC via Bayesian Set Optimization for Dynamic Manipulation of Deformable Linear Objects

cs.RO ·2026-10-22 ·Youyuan Long, Gokhan Solak, Arash Ajoudani
code/data
Abstract

Dynamic manipulation of deformable linear objects (DLOs) remains challenging due to underactuated, high-dimensional dynamics, delayed deformation responses, and the presence of multiple feasible motions under obstacle constraints. This paper presents BM2PC, a Bayesian Multimodal MPC framework for agile manipulation of DLOs. Building on the control-as-inference posterior, BM2PC formulates multimodal planning as a set optimization problem that balances control quality and behavioral diversity. An efficient incremental update maintains multiple promising control modes with low additional computational overhead. For real-world deployment, BM2PC incorporates an execution dynamics model that predicts actual end-effector motion and compensates for execution mismatch during fast manipulation. We evaluate BM2PC on rope tracking and obstacle avoidance in simulation and hardware, and on simulated disturbance rejection. In trajectory tracking, BM2PC generalizes better to unseen trajectories than a learning-based baseline without task-specific policy training or fine-tuning. Real-world experiments demonstrate accurate long-horizon tracking, reducing RMSE by 44.0% on the longest task. In obstacle avoidance, BM2PC achieves comparable performance to existing multimodal MPC methods while reducing planning time by 29% to 66%. These results demonstrate accurate dynamic control and efficient multimodal planning within a unified MPC framework. The core implementation of BM2PC and supplementary videos are available at https://github.com/Longyouyuan/BM2PC.

0.22

OpenTac: An Open Ecosystem for Customizable, Scalable, and Transferable 3D Tactile Sensing

cs.RO eess.SPeess.SY ·2026-10-22 ·Ariel Slepyan, Geeta Chandra Raju Bethala, Jiafeng Gu, Anagha Rajendra Dangle, Yinzhe Zhou +10 more
code/data
Abstract · 15 pages, 15 figures. Project website, open-source hardware and software, and demonstration videos: https://opentac.github.io

We present OpenTac, a novel sensor and a full-stack ecosystem for tactile skins that are highly customizable, scalable, and readily deployed in downstream robot control and learning applications. Our core contribution is twofold. First, we develop a scalable resistive tactile skin with a structured deformable overlay that converts tangential loading into spatially varying pressure. Differential responses across neighboring "taxels" support decoding of local shear vectors, and the common-mode response encodes normal force. Second, we provide a highly customizable sensor design tool that is able to automatically generate fabrication files for these tactile skins from 2D outlines and developable 3D surfaces, including from abstract descriptions such as text. We develop an ecosystem around the OpenTac sensor, including the CAD design tool, a sensor simulator, cross-sensor transfer, and fast/adaptive data acquisition. We demonstrate the broad applicability of the OpenTac-CAD sensor design tool by designing and fabricating more than 30 sensors across a variety of robot embodiments. We characterize the three-dimensional tactile fields produced by OpenTac with a commercial, high-precision force/torque sensor. We demonstrate the broad applicability of 3D tactile sensing across contact-rich feedback control and manipulation policy learning on four real-world robot embodiments. Additionally, adaptive acquisition enables kilohertz-rate partial readouts, and our readout circuitry enables high dynamic range (HDR) acquisition, measuring, for example, a 3.7 kg weight stack and a 20 g weight on the same sensor in the same frame. We provide all hardware, software, and result videos at opentac.github.io

0.21

Full-Weight VLA Fine-Tuning Without an fp32 Master Copy Silently Erases Most Updates

cs.RO ·2026-10-22 ·Shubham Singh
Abstract

When a vision-language-action (VLA) model is fine-tuned with its trainable weights in bfloat16 (bf16) and no fp32 master copy, AdamW writes each update straight into bf16, silently rounding away any update smaller than 1/512 of a weight. In 9 of 12 open VLA codebases a default or documented recipe trains this way, 4 on full weight matrices, including LeRobot's SmolVLA default and its $π_0$/$π_{0.5}$ guides. In our LIBERO runs without the copy, a probe every 100 steps finds at least 79% of updates to bf16 weights erased. On the Hugging Face Hub, 99% of sampled full-weight SmolVLA fine-tunes store the action expert in bf16, and 84% of recent $π_{0.5}$ fine-tunes with a training configuration set the bf16 flag. On LIBERO-Object, with rules fixed in advance, an fp32 copy raises success over LeRobot's storage from 4.3% to 49.3% for SmolVLA and from 5.1% to 80.9% for $π_{0.5}$ at a ten-times lower learning rate, where both models are undertrained, and from 89.0% to 97.7% for $π_{0.5}$ at its default rate. For SmolVLA at its default rate it adds 10.2 points with open-loop 50-action chunks but only 4.2 when re-planning every 10 actions. In these five comparisons and two more against SmolVLA with every trainable weight in bf16, every run with the copy beats every run without it (p < 0.05); all gaps but the 8.7- and 4.2-point ones clear our 10-point threshold. The fix is to keep the trainable weights in fp32 during training, a one-line change for SmolVLA in LeRobot.

0.21

CoHelix: Event-Based Multi-Propeller Rotation-Rate Tracking with Shared Projective Geometry

cs.RO ·2026-10-22 ·Radim Spetlik, Tobiáš Vinklárek, Jan Klivan, Tomáš Kestřánek, Matouš Vrba +1 more
Abstract

Estimating propeller rates from a moving event camera requires separating blade rotation from the changing image geometry. Background activity mixes with blade events, while repeated blade appearances admit half- and double-rate ambiguities. We propose CoHelix, a recursive filter that estimates one shared homography and separate rotation states for four rigidly mounted, coplanar propellers. Each admitted event updates its assigned rotation state, and its accepted geometry measurements correct the shared homography used to interpret later events from every propeller. We report lock, the fraction of evaluated time for which an estimate remains within 2% of synchronized ground truth, because mean error alone does not show how long a tracker follows the true rate. With fixed-rate initialization, leave-one-recording-out evaluation on moving-camera data gives CoHelix 62.9% lock, compared with 39.8% for HelixTrack and 22.2% for EventPro. With other settings fixed, allowing four geometry contributors instead of one raises lock from 3.4% to 62.9%. On a private target-follower dataset with a different target quadcopter, CoHelix retains the measurement settings selected on moving-camera data and uses a fixed initial rate. It reaches 95.9% lock, compared with 56.2% for HelixTrack and 61.1% for EventPro.

0.21

See What I Need, Dream What I Want: A Navigation Foundation Model on a Self-Constructed Representation Space

cs.RO cs.CV ·2026-10-22 ·Jiawei Chen, Yinxi Wang, Zihan Yang, Meijun Gao, Yaowei Liu +10 more
code/data
Abstract · 28 pages. Project page: https://wang-yinxi.github.io/representnav/

What visual representation should a navigation policy act on? Navigation policies built on vision foundation models largely leave representation design to the encoders' pretraining objectives. We present RepresentNav, a navigation foundation model built on a self-constructed representation space: a visual representation constructed by the policy's own objectives from vision foundation models. RepresentNav builds this space with Representation Residuals, a compact three-layer transformer that attends over features spanning the full depth of multiple complementary vision foundation models, so that the policy learns to see what it needs for navigation. In the same space it dreams what it wants: before every action a latent chain-of-thought predicts the representation of the scene a moment ahead. The dream is trained end-to-end with the policy as a joint-embedding predictive architecture (JEPA), with a next-representation prediction loss and a regularizer on its targets. RepresentNav sets a new state of the art on VLNVerse, VLN-CE R2R, EVT-Bench, and two point-goal benchmarks, and its gains carry over to a physical robot amid clutter and moving people. On VLNVerse fine-grained, RepresentNav achieves 83.80% success, outperforming the previous state of the art by 20.05 percentage points, and on EVT-Bench distracted tracking it reaches 85.5% with a single camera (+10.1). With the same encoders, the self-constructed space clearly outperforms concatenating their outputs and learns to weight their layers unevenly, and a dream in it cuts collisions by a quarter. A single dreamed latent decodes into the depth, semantics, and 3D occupancy of the scene a moment ahead, more accurately than pretrained encoders applied to the real future frame, although occupancy is never supervised, and the rollout tracks the robot's own motion over the next second: the dream emerges as a compact latent world model.

0.21

Whole-Body Motion Guided Reference Generation for Uncertainty-Aware Compliant Assistance

cs.RO ·2026-10-22 ·Muyuan Ma, Long Cheng, Lijun Han, Haotian Zhai, Xinpan Meng +3 more
Abstract · Withdrawal requested because not all co-authors had consented to the submission of this manuscript to arXiv

Exoskeletons have shown great potential for augmenting human physical capabilities, yet generating reliable assistance remains challenging because human motion patterns and interaction requirements vary substantially in practice. To address this problem, a dual-loop assistance strategy is proposed to couple multimodal exoskeleton trajectory generation with variable admittance control, where trajectory variance links the outer-loop generator and the inner-loop controller. The generator synthesizes future exoskeleton trajectories through denoising, primarily using whole-body human motion and exoskeleton proprioception, with task descriptions as auxiliary semantic cues. To reduce cross-modal representation mismatch, a three-stage training pipeline is introduced to progressively optimize the human motion encoder, the exoskeleton proprioception encoder, and the trajectory generator. In the inner loop, multiple future trajectories are sampled from the generator to estimate the trajectory mean and variance. The mean is used as the control reference, while the variance provides an uncertainty measure for prediction reliability. This measure is further incorporated into a stiffness regulation law, enabling compliant assistance with input-to-state stability under bounded lumped interaction input. Experimental results demonstrate that the proposed method significantly reduces muscle activation levels compared with standard baselines, while also decreasing users' metabolic energy expenditure relative to the unassisted condition.

0.21

Exposing Noise-Deficiency Signatures of Deep Learning-Based False Data Injection Attacks Through AC State Estimation

eess.SY ·2026-10-22 ·Yichen Liu, Hongyu Wu, Bo Liu
Abstract

Deep learning (DL)-based false data injection (FDI) attacks can resemble normal measurements and evade both conventional bad data detection (BDD) and measurement-space data-driven detectors. This paper identifies a previously overlooked signature of such attacks in state-estimation (SE) residual space. The residuals of reconstruction-based DL attacks exhibit deficient stochastic variation, manifested as abnormally small residual energy and reduced directional dispersion. We establish the conditions under which this signature emerges and explain why the attacks carrying it can evade conventional BDD. We show that SE can reveal residual-space differences obscured by operating-state variation in measurement space. We then explain the origin of the signature through the noise-suppression mechanism of DL attack generation. A generator can learn the reproducible structure of historical measurements but cannot reproduce stochastic measurement components that are unpredictable from its input. This signature enables label-free detection using a simple two-sided test on the conventional BDD statistic, while the full residual vector can also be exploited when labeled attack samples are available. Experiments on the IEEE 14-bus and 118-bus AC systems show that the label-free test detects up to 99% of three representative DL attacks at a false-alarm rate comparable to that of BDD, which detects at most 5%. Ablation studies further examine the effects of residual representation, decision rule, and measurement-noise magnitude and distribution.

0.21

Consensus-Based Distributed Filtering for Angles-Only Tracking in Cislunar Space

eess.SY cs.MA ·2026-10-22 ·Giovanni Romagnoli, Fabrizio Pilone, Giordana Bucchioni
publishedbenchmark
Abstract · Accepted for presentation at IEEE NAVICON 2026

This paper investigates consensus-based distributed filtering strategies for multi-observer angles-only navigation in the cislunar space. The objective is to estimate the inertial state of a target spacecraft using only line-of-sight measurements collected by a network of observer spacecraft. This problem is particularly relevant for autonomous cislunar navigation, where ground support may be limited, and centralized fusion architectures may be problematic due to communication and computational constraints, and single-point failure risks. Three distributed filtering approaches are considered: the Distributed Extended Kalman Filter, the Information-Weighted Consensus Filter, and the Kalman Consensus Filter. All methods share the same local prediction and measurement-processing steps, but differ in the way information is exchanged and fused across the network. Their performance is evaluated and compared to a centralized Extended Kalman Filter, adopted as benchmark, and independent local filters in a high-fidelity cislunar scenario. Numerical simulations show that the distributed approaches significantly improve estimation accuracy with respect to independent filters. Monte Carlo sensitivity analyses are then conducted to characterize the robustness of the considered methods to finite consensus iterations and prior misalignment among the agents.

0.21

Live Thread Reconstruction for Autonomous Manipulation

cs.RO ·2026-10-22 ·Emma Huang, Omar Hernandez, Zih-Yun Chiu, Michael Yip
Abstract

Autonomous robotic surgery has drawn increased attention in recent years due to its potential to relieve surgeons of time-consuming, mentally straining, and tedious tasks, particularly during prolonged operations. One suitable task for automation is suturing, which involves tracking the thread as it deforms during needle retrieval, de-tangling, knot tightening, and other tasks. Unlike rigid-object manipulation, deformable thread manipulation has no fixed pose to track and effectively infinite degrees of freedom, making its configuration difficult to predict and requiring the thread's state to be re-estimated rather than assumed. At the same time, thread manipulations involve sequential motions that depend on past and current thread contact points, which must be accurately and continuously tracked. This paper presents an image-to-manipulation pipeline that concurrently generates live thread reconstruction while conducting safe and optimized sequential manipulations to achieve long-horizon tasks.

0.21

VecTile: A Low-Cost Flexible Tactile Array for 3-DoF Force Sensing

cs.RO ·2026-10-22 ·Jian Hou, Xin Zhou, Adam J. Spiers
Abstract · 8 pages, 8 figures. Submitted to IEEE International Conference on Robotics and Automation (ICRA) 2027

Tactile sensing provides robots with contact information difficult to obtain from vision alone. Shear force is particularly important for understanding tangential contact, guiding corrective motions, and maintaining grasp stability, yet flexible and low-cost tactile arrays often measure only normal force. We introduce VecTile, a low-cost, 6 mm thin, flexible tactile array capable of three-axis force sensing. VecTile combines a piezoresistive layer with passive silicone nubs that convert normal and shear forces into distinguishable pressure patterns. The sensor can be fabricated in-house without specialized equipment. The complete system costs less than 35 USD for a 4x4 array. We validate the sensing mechanism through finite element analysis and hardware experiments. In controlled single-taxel tests over 1-5 N normal force and up to 1.5 N shear force, VecTile achieves root-mean-square errors of 0.39 N and 0.16 N for normal and shear forces. A VecTile calibrated while flat retains accuracy when curved, while a five-minute manual calibration is sufficient for practical use. Two applications demonstrate its utility. First, shear readings are sensitive enough to stop a sharp scalpel from puncturing a balloon after contact. Second, the sensor substantially improves Diffusion Policy-guided power-plug insertion. All designs and code will be released upon paper acceptance. Project website: https://jiannnh.github.io/VecTile/.

0.20

On Sample Complexity Bounds for Frequency-Domain System Identification

eess.SY ·2026-10-22 ·Arya Honarpisheh, Mario Sznaier
theory
Abstract

We study the sample complexity of frequency-domain system identification with bounded measurement noise. We focus on interpolatory identification algorithms, which return models consistent with both the data and a priori constraints. For equally spaced samples, we sharpen previous worst-case bounds on the consistency-set radius. Its excess above the noise level decays as $N^{-2}$, where $N$ is the number of samples, and a matching lower bound shows that this rate is tight. We then examine whether bounded stochastic noise allows identification below the noise level. We show that high-probability error bounds can converge to zero. Their convergence rate depends on how strongly the noise distribution is concentrated near the boundary of the noise set. For example, noise uniformly distributed on a disk gives an upper bound of order $(\log N/N)^{2/3}$. Numerical experiments examine these convergence rates and the effect of the noise distribution on the identification error.

0.20

When One Step Is Enough: An Empirical Study of Euler Integration in Flow-Matching Policies

cs.RO ·2026-10-22 ·Shivakanth Sujit, Luca Nunziante, Rousslan Fernand Julien Dossa, Kai Arulkumaran
published
Abstract · Accepted at the 10th Conference on Robot Learning (CoRL 2026)

Flow matching is now the dominant approach for vision-language action (VLA) robot policies, thanks to its expressiveness and simple training objective. However, inference for standard flow matching requires iteratively solving an ODE, limiting the control frequency at deployment. While several works have proposed novel methods to allow for single-shot action generation, we ask whether such additional objectives are actually necessary. We find that across six flow matching VLAs, one-step action generation largely matches multi-step sampling at scale: multi-step integration confers little to no consistent advantage at inference. We empirically find that this holds as long as models are sufficiently trained and have enough capacity. Practically, the implication is that for large VLAs, the single-step policy comes essentially for free from the multi-step model, enabling low-latency deployment without retraining, distillation, or auxiliary objectives.

0.20

Learning Robot Behavior Priors from Non-Robotic Synthetic Data

cs.RO cs.LG ·2026-10-22 ·Mateusz Odrowaz-Sypniewski, Michael Amir, Jasmine Bayrooti, Ajay Shankar, Amanda Prorok
Abstract

Real robot data, rich across tasks and morphologies, is often treated as the bottleneck for training robot learning models, including embodied foundation models that aim for zero-shot physical transfer. In this paper, we investigate the degree to which such models can acquire necessary structure using data from systems without embodiment. To this end, we test if inexpensive, procedurally generated synthetic data can yield useful priors before a model observes any real-world behavior. We study this through robot action prediction, a natural pretraining objective and a direct test of whether structure learned from arbitrary synthetic tasks transfers to embodied settings. We introduce three data generation pipelines that model mathematical abstractions of robotic control, and use them as pretraining priors for a Transformer. We demonstrate that a model trained purely on this non-robotic data attains non-trivial zero-shot prediction performance on unseen physical robot tasks via an in-context learning mechanism, and that our specialized control priors notably improve upon generic causal priors. We also compare against a compute-matched baseline trained on real robot data, and find that it outperforms our model but only on the datasets it was trained on. These findings suggest that procedurally generated, mathematical abstractions can yield useful robotic priors, offering a potential complement to expensive embodied data collection.

0.20

Multi-Tool Selection for Robotic Systems: Comparing Semantic Retrieval and LLM

cs.RO cs.AI ·2026-10-22 ·Miguel Sousa, Luis Ramos Pinto, João P. Matos-Carvalho
Abstract

Natural-language interfaces allow users to command robotic systems without interacting directly with mission planners or robotics middleware. Their reliable use depends on robot tool routing, defined here as the selection of the smallest complete subset of predefined executable capabilities required to satisfy a request. This task is particularly difficult for compound instructions, where incomplete selection may omit necessary sensing or monitoring capabilities, while excessive selection may expose irrelevant tools. We compare four routing strategies: three progressively refined ChromaDB-based methods - fixed-threshold retrieval, relative-distance retrieval, and clause-aware aggregated retrieval - and a cloud-hosted 120-billion-parameter language-model baseline. The methods are evaluated on 164 manually annotated prompts organised into three complexity levels, from direct single-tool requests to compound requests requiring at least three tools. Using exact-match accuracy, completeness, precision, a completeness-weighted composite score, and routing latency, CDB-v3 achieves the best macro-average score among the retrieval-only methods, with 0.7822, compared with 0.8869 for the large-model baseline. Although the LLM provides higher routing quality, CDB-v3 exhibits substantially lower latency and can be deployed locally without requiring cloud connectivity. These results identify a quality-latency trade-off between direct large-model selection and clause-aware semantic retrieval for privacy-sensitive, connectivity-constrained, and resource-constrained robotic systems.

0.19

Cross-Region Cooperation of UAV Swarms Towards Air-Ground ISAC via Hierarchical Multi-Agent Reinforcement Learning

eess.SY ·2026-10-22 ·Linghui Miao, Nhan Thanh Nguyen, Shijian Gao
Abstract · 13 pages, 12 figures

The limited communication resources and sensing perspectives of a single uncrewed aerial vehicle (UAV) motivate UAV swarms for integrated sensing and communication (ISAC) in air-ground networks. However, in multi-region scenarios, regional UAV allocations may not match vehicle demand, while local observations may lack sufficient diversity for accurate sensing. To address these issues, this paper proposes a cross-region cooperation framework for UAV swarms. For communication, cross-region UAV reallocation reduces service disparities by matching resources to vehicle demand. For sensing, combining UAV observations from different regions increases observation diversity, while adaptive handshaking guided by marginal Cramér--Rao bound (CRB) utilities balances phase synchronization gains against handshaking energy consumption. Joint optimization of UAV positions and transmit power within each region then further improves communication service and sensing performance. These coupled decisions are coordinated by a hierarchical multi-agent reinforcement learning (H-MARL) approach through slow-timescale inter-region UAV reallocation and fast-timescale intra-region UAV control. Under the considered simulation settings, reallocation can improve quality of service (QoS) by over 20 percentage points compared with retaining the initial allocation under severe resource-demand mismatches. Cooperative sensing with adaptive handshaking incurs only about a 5.0% increase in mean CRB relative to perfect synchronization.

0.19

Know When to Escalate: Critic-Guided Adaptive Execution for Cloud-Local Embodied Agents

cs.RO cs.DC ·2026-10-22 ·Wei Zhang, Lirong Xiang, Charles Fleming
Abstract

Recent advances in foundation-model agents have enabled increasingly capable embodied systems for long-horizon planning and interaction. However, embodied agents require fast, closed-loop decisions as they continuously perceive and act on a changing environment. Deploying strong cloud agents leaves the platform idle for seconds between consecutive actions, while lightweight local models respond quickly but are more vulnerable to errors. Existing hierarchical architectures typically assign models fixed roles, whereas conventional model routers make stateless per-request decisions. We present CARE, a critic-guided adaptive execution framework that dynamically determines when to escalate control from a local executor to a stronger cloud model. A lightweight critic model monitors execution reliability and triggers escalation when local execution becomes unreliable; once escalated, control persists until the agent recovers, avoiding step-wise switching. To support such control transfer, a shared handoff memory maintains execution context and task-level progress across executors, enabling the model to intervene without restarting from scratch. Across diverse embodied benchmarks, CARE improves task success by 6.3-18.3 percentage points over the strongest evaluated collaboration baseline. CARE further approaches or exceeds Cloud-only success while reducing median time-between-action latency by 28.8-82.6%.

0.19

Humanoid Mobility via Attention-based Contacts

cs.RO ·2026-10-22 ·Junzhe He, Kehan Wen, Chong Zhang, Clemens Schwarke, Zewei Zhang +5 more
Abstract · under review

Humanoid mobility over complex terrain requires accurate foot placement and coordinated whole-body motion capable of robust adaptation. Reinforcement learning (RL) policies trained without motion references generalize well across terrains, yet eliciting effective whole-body coordination through reward design and exploration alone remains challenging. In contrast, reference-based methods yield natural motions but struggle to adapt beyond the predefined trajectories in their training data. We introduce a learning framework that bridges these two approaches by using motion data strictly as a training prior. An explicit, self-supervised, attention-based contact planner predicts future body-environment contacts, while a contact-conditioned tracker translates these plans into coordinated whole-body motion. Together, they establish contact as an interface between sensory observations and control actions. This abstraction allows the policy to internalize the contact-aware structure of locomotion rather than directly replicating kinematics. Consequently, a single policy can exploit data-driven motion priors for natural whole-body coordination while retaining the flexibility to discover environment-dependent behaviors through reinforcement learning. Building on the learned contact planner and tracker, a subsequent fine-tuning stage bootstraps their acquired capabilities to explore environment variations beyond those covered by the original motion data. Operating entirely from onboard observations without any reference inputs at deployment, this framework enables versatile humanoid mobility in challenging environments.

0.19

Co^2Gen: Collaboratively Generating Task Plans for Decentralized Multi-Robot and Human Collaboration

cs.RO ·2026-10-22 ·Yi Loo, Siddhant Shrivastava, Minh Long Pham, Malika Meghjani
Abstract

Visual Language Action (VLA) models are designed to be generalist robot models that can be implemented across embodiments, tasks and environments. However, their ability to collaborate with other robots and humans remains underexplored. In this work, we propose a framework to automate the generation of multi-robot collaborative data for decentralized VLA training without requiring human teleoperation. Our framework first utilizes a decentralized collaborative planning approach to generate multi-robot collaborative robot plans in the form of an executable function. We then utilize the collaborative plan to produce collaborative demonstration data to finetune VLA models in a decentralized manner. We show that our approach produces demonstration data that when used to finetune VLA models on decentralized collaboration tasks matches the task performance of VLA models finetuned on human-teleoperated collaborative data.

0.19

TRACE: Traversability-aware Reactive Navigation via VLM-driven Adaptive Constraints Estimation

cs.RO ·2026-10-22 ·Mario A. V. Saucedo, Midhun Edatharam Kunnath, Elias Small, Akash Patel, Shruti Kotpalliwar +2 more
published
Abstract · Presented at IFAC 2026

This paper presents TRACE, a traversability-aware reactive navigation framework designed for the autonomous operation of ground robots in unstructured forest environments. Unlike traditional approaches, TRACE leverages VLM-based confidence maps from positive and negative natural-language prompt sets and combines them through evidential fusion to produce a dense traversability map that prioritizes the terrain based on signs of previously traversed paths. Traversability-annotated points are then extracted and used to derive the left and right boundaries of the traversable path via percentile binning followed by spline or regularized polynomial fitting. These boundaries are then incorporated as soft barrier constraints within an NMPC formulation, enabling safe and adaptive path following in complex natural environments. Rigorous experimental trials carried out in real-life forest environments demonstrate TRACE ability for smooth navigation in cluttered, deformable, and visually ambiguous terrains.

0.18

In-Orbit Computing for Remote Sensing: Task-Oriented Processing and Transmission

eess.SP ·2026-10-22 ·Yi Zhang, Zesong Fei, Jingxuan Huang, Hua Dang, Jihao Luo +3 more
Abstract

The growing volume and timeliness requirements of remote sensing data are exposing the limitations of the conventional "transmit-then-process" architecture, particularly under intermittent satellite-ground connectivity, constrained downlink bandwidth, and limited onboard computing resources. This article presents a semantic-aware space-ground collaborative remote sensing (SSGC-RS) framework, in which remote sensing data are processed to different semantic depths according to task requirements and available communication-computing resources. Three operating modes are considered, ranging from fully onboard task inference to semantic representation delivery and space-ground split inference. A land-cover recognition case study illustrates how these modes address different communication and computing bottlenecks. Finally, emerging challenges and future directions for intelligent remote sensing systems are discussed.

0.18

History-Aware Control Barrier Functionals with Nonsingular Fading-Memory Kernels

eess.SY ·2026-10-22 ·Ege C. Altunkaya, Esra Demir, Murad Abu-Khalaf, İbrahim Özkol
benchmark
Abstract

Safety-critical protection often depends not only on the state and physical limits of the system, but on how that state evolved over time. Here, the safety margin measures the remaining allowance before a physical limit is reached. In this context, the objective is to enforce both physical safety and history-dependent retention of this margin for integer-order control-affine systems. To this end, changes in this margin are converted into memory through fading-memory kernels with finite nonnegative weights. A history-aware control barrier functional (HA-CBFal) is constructed explicitly, with nonnegativity equivalent to retaining a prescribed fraction of this margin's weighted historical value. This condition is enforced by an affine input constraint and proved to imply physical safety under stated kernel, initialization, and feasibility conditions, without finite-dimensional memory realization. In 36 calibration cases on the ADMIRE flight-control benchmark, mean per-case tracking-RMSE reductions of 27.2\% and 7.0\% are obtained relative to calibrated state-only CBF and moving-average control barrier functional (MA-CBFal), respectively, with both constraints preserved to numerical tolerance. In a shorter-recovery maneuver outside calibration with frozen gains, both constraints are preserved to tolerance by HA-CBFal and state-only CBF, but the retention requirement is violated despite MA-CBFal nonnegativity.

0.18

Causal Analysis of Failures of Cyber-Physical Systems with Continuous Inputs

cs.RO ·2026-10-22 ·Tom Yaacov, David A. Kelly, Khen Elimelech, Hana Chockler
Abstract

Modern autonomous Cyber-Physical Systems (CPSs), such as self-driving cars, operate in increasingly complex environments while expected to maintain high reliability. The black-box nature of many of these systems makes it impossible to fully verify their behavior prior to deployment. When a failure occurs, it is important to analyze it to guide improvements to system robustness. A recent work introduced the novel concept of leveraging the framework of actual causality for CPS failure explanation. However, that work considered only a discrete, binary causal model that cannot capture continuous changes in system inputs. In this paper, we address these limitations by extending the actual causality framework with the notion of distance to account for the continuous characteristics of the system. Specifically, we introduce the novel concepts of distance-weighted actual cause and distance-weighted degree of responsibility, enabling causal influence to account for the magnitude of changes required to avoid a failure, rather than only its cardinality. While motivated by the continuous nature of CPSs, this constitutes a general extension of the theory of actual causality and enables more fine-grained analysis of continuous inputs in any causal model. Building on these notions, we present an efficient, practical, system-agnostic explanation-derivation algorithm that uses distance-weighted responsibility to guide the explanation search. We experimentally evaluate our approach across a number of environments from two autonomous-driving simulators with black-box neural controllers. The results show that the proposed extension provides more effective guidance than the current discrete approach.

0.17

WorldMirror: Agentic Real-to-Sim Generation of Interactive Worlds from In-the-Wild Robot Demonstrations

cs.RO cs.CV ·2026-10-22 ·Tianling Xu, Weijie Wan, Honghao Su, Kailun Su, Junyuan Tang +7 more
Abstract

Existing approaches for robot learning in simulation often require substantial manual effort to construct scenes and specify tasks, limiting training scalability and data diversity. Meanwhile, in-the-wild robot demonstrations naturally provide diverse and realistic observations of real-world interactions. Leveraging these demonstrations for simulation requires both high-fidelity visual reconstruction and physically consistent interactions. General-purpose agents offer flexible adaptation to diverse scenarios and strong reasoning about 3D spatial relationships, while specialized generative models provide high-fidelity reconstruction of scene appearance and geometry. We introduce WorldMirror, an agentic real-to-sim framework that constructs interactive digital twins from in-the-wild robot demonstrations. Our central idea is to formulate real-to-sim construction as the iterative refinement of an executable world program. A coding agent decomposes the observed scene, plans reconstruction steps, and orchestrates specialized image and 3D generation tools, refining the program through verification feedback. A camera-pose-anchored scene representation organizes the world into static visual context, support geometry, and editable interactive assets while preserving spatial relationships among the scene, objects, and robot. The agent uses this representation to diagnose and correct reconstruction errors, enabling physically consistent interaction and trajectory replay. Experiments on DROID demonstrate that WorldMirror can reconstruct visually faithful and geometrically consistent interactive worlds while preserving the physical validity of recorded robot trajectories.

0.17

Incentive Mechanisms for Welfare Maximization Under Individual Rationality in Noncooperative Dynamical Systems

eess.SY ·2026-10-22 ·Yuyue Yan, Maojiao Ye, Hideaki Ishii, Tomohisa Hayakawa
Abstract · 16 pages, 10 figures

Human societies are shaped by strategic interactions among self-interested decision-makers, where conflicts between individual and social interests naturally arise. This paper investigates the design of incentive mechanism for multi-agent game systems to reconcile social welfare maximization with individual benefits under economic constraints. We develop a linear-quadratic incentive framework to reshape Nash equilibria while reconciling the constraints of Pareto efficiency, individual rationality, and budget balance. The optimal target state is identified as the solution of a constrained optimization problem associated with the social welfare and aggregated payoff functions. Our theoretical analysis characterizes the conditions of incentive functions that steer self-interested agents towards the optimal target state as a Pareto-efficient Nash equilibrium for achieving the maximum attainable social welfare. We further characterize how the choice of social welfare function constrains achievable individual outcomes, revealing a structural paradox associated with priority weighting under individual rationality. Several numerical examples are presented to demonstrate the efficacy of our results.

0.17

SR-ANM: A Square-Root Atomic Norm Minimization Estimator for Noise-Robust Direction-of-Arrival Estimation

eess.SP ·2026-10-22 ·Youval Klioui
Abstract

This paper proposes a Square-root Atomic Norm Minimization (SR-ANM) estimator for direction-of-arrival (DoA) estimation in the single measurement vector (SMV) setting. This new framework, inspired by the Square-root Least Absolute Shrinkage and Selection Operator (Square-root LASSO), relies on a residual term based on the $L_{2}$-norm rather than the usual squared $L_{2}$-norm so as to obtain an estimator whereby the hyperparameter of the atomic norm in the unconstrained convex optimization problem exhibits robustness to the noise level compared to classical ANM. Experimental verification across a wide signal-to-noise ratio (SNR) range from 0 dB to 40 dB using the the probability of detection and the angular root-mean-squared error (RMSE) confirms the increased noise robustness of the proposed framework compared to ANM.

0.17

Robotic Budding: From Homogeneous Cells to Diverse Morphologies in Physical Self-Reproduction

cs.RO ·2026-10-22 ·Siyuan Zhang, Junyan Liu, Wanglin He, Hod Lipson
Abstract

Physical self-reproduction in modular robots remains difficult because a robot must not only move, but also collect, align, dock, and separate modules reliably in hardware. Taking inspiration from how genetically identical cells form the varied tissues of an organism, we introduce a single standardized robotic module that combines a magnetic docking interface with two actuated joints and folds from a flat, two-dimensional configuration into three-dimensional structures. Modules of this single design can be assembled into two morphologies: a three-module "trilobite" and a five-module "seal". Using a small set of discrete motion commands, a human operator steers an assembled parent robot to dock magnetically with a sequence of spare modules, which are assembled into a second, structurally identical individual that then separates from the parent. Because the parent retains all of its original modules, this is a budding-style form of physical reproduction. Here, we report preliminary hardware demonstrations of this full process on both morphologies. We discuss the current scope of the system, including its present reliance on direct human teleoperation rather than autonomous control, and outline directions toward autonomous execution of this process and repeated validation across a larger number of independent trials.

0.17

EBV-CAN: Vision--Language Grounding for Contact-Aided Endoscope Navigation

cs.RO ·2026-10-22 ·Chi Kit Ng, Shuchang Liu, Jinsong Lin, Yidong Zhang, Xizi Wang +2 more
Abstract

Flexible endoscopes exploit structural compliance to traverse narrow lumens, pipes, and cavities, where navigation involves distributed environmental contact. This compliance makes tip motion depend on the evolving support configuration. We present external bird's-eye-view contact-aided navigation (EBV-CAN), a vision-language policy that combines one initial global static bird's-eye image with evolving endoscopic observations and motor state. Three supervised semantic queries locate the tip, next waypoint, and instructed target on the cached global image; their residual representations predict bending and insertion commands. We introduce the Vision-Language Contact-Aided (VLCA) dataset, linking 517 demonstrations and 75k action samples to 958k spatial entries and 93k remaining-route records. On 50 held-out episodes, tip-coordinate and waypoint localization reach 97.5% and 94.8%. Swapping the target instruction under identical observations changes target attention and motor output across 17 dual-target episodes. Across 1,920 physical trials of eight policies trained under a shared protocol, EBV-CAN achieves 183/240 successes (76.25%) under automatic blue-area scoring. Gated recurrent unit behavior cloning (GRU-BC) achieves the lowest matched-task offline mean absolute error (MAE), 0.056, and 1/240 physical successes, establishing distinct rankings for action imitation and task completion. Successful multi-guide executions and representative failures connect spatial progress, steering transitions, and visibility to physical navigation.

0.16

Barrier Function Control for the Interconnection of DC Electronic Loads to Utility Grid

eess.SY ·2026-10-22 ·Grigoris Michos, Christos K. Verginis
Abstract · preprint for paper to appear in IEEE ISGT Europe conference 2026

This paper proposes a nonlinear control law that facilitates the safe integration of DC electronic loads to the utility grid. Specifically, by employing suitable reciprocal barrier functions, it is rigorously proven that the closed-loop system guarantees constrained operation within a desired operating region, despite the influence of fast converter dynamics and fluctuating constant power load demand. This way, the proposed controller guarantees safety of the electronic components from surges in power demand and high transients. The theoretic results are validated in simulation scenarios, where the performance of the proposed controller is compared against the conventional saturated-PI controller. Contrary to the latter, the proposed method guarantees stable and safe operation, enhancing the system robustness.

0.16

Continuous-Time Motion Planning and Steering with Global Convergence

cs.RO ·2026-10-22 ·Mohsen Amiri, Sima Didari, Davoud Ataee Tarzanagh, Hankyu Moon, Yeongdae Kwon
benchmark
Abstract

Motion planning is central to robotics, and it is increasingly used to correct the motions proposed by learned planners and policies. In both uses, constraints must hold at every instant of the motion, not only at sampled times. To our knowledge, no algorithm provably converges to a globally optimal motion once a smoothness cost is combined with such constraints. We cast planning and steering as a single continuous-time program: robots avoid one another and static or moving obstacles, and a learned proposal is steered by keeping the motion close to it. The program is elementary to state, yet nonconvex. We propose GlobMP (Global Motion Planning), a gradient projection method over piecewise-polynomial curves, on which each constraint is one polynomial per piece that must stay nonnegative over the whole motion; it alternates cost reduction with feasibility restoration. Under mild assumptions, we show that, from a fine grid start, its cost approaches the global optimum as the number of polynomial pieces grows, with every constraint met to any chosen tolerance, or exactly on stable instances. We demonstrate its superior performance over strong baselines in real and synthetic experiments, including teams of up to $200$ robots, multi-robot planning on the Robotarium testbed and an extended benchmark, maze navigation, and steering tactile-reactive policies on a bimanual humanoid.

0.16

Calibration-free IMU-based Relative Orientation Estimation under Dynamic Human Motion

cs.RO eess.SY ·2026-10-22 ·Ruiyuan Li, Hassan Osman, Ajay Seth, Manon Kok
Abstract · 8 pages, 7 figures, 1 table. Submitted to IEEE Sensors Journal

Recent magnetometer-free inertial measurement unit (IMU) methods can reduce drift in joint motion estimation using rigid-body constraints, but they typically require pre-calibrated sensor locations, assumptions of specific joint models (e.g., hinge joints) or limited motion types, or an external reference for global orientation. Furthermore, challenges related to high-frequency impacts (e.g. at foot strikes during gait) and numerical differentiation of noisy measurements have not been explicitly addressed. We propose an extended Kalman filter that estimates relative IMU orientation and sensor location simultaneously, requiring neither offline sensor location calibration nor a global orientation anchor. A systematic simulation study shows that the quality of the relative orientation and the sensor location estimates depends on different motion-intensity sources: the joint center acceleration and the relative angular velocity, respectively. A low-pass filter (LPF) is used to mitigate real-signal outliers before applying the rigid-body constraint. When applied to the OpenSense dataset with real IMU measurements across four subjects with three lower-body joint types during walking and complex tasks including various motion types, the proposed method achieves an overall relative orientation error of 2.61°, compared with 10.40° for a baseline using offline-estimated sensor locations and 7.43° for a baseline using marker-derived sensor locations. These results show that the proposed calibration-free estimation framework handles real IMU data effectively and achieves accurate relative orientation without offline sensor location calibration.

0.16

Feeder-Aware Service-Conditioned Flexibility of Grid-Interactive Buildings

eess.SY ·2026-10-22 ·Dafang Zhao, Ittetsu Taniguchi, Takao Onoye
Abstract

Grid-interactive buildings can adjust demand, but building-feasible power budgets may differ from accepted feeder-head offers. This study addresses a practical question: to what extent can the adjustable electricity consumption of buildings be translated into deliverable flexibility services while satisfying feeder-head operating requirements? To enable a fair comparison, we adopt the same baseline operating schedule, equipment conditions, and post-service recovery requirements for indoor temperature and battery state of charge (SOC). Flexibility is then evaluated under two settings: one considering only building-level constraints, and the other additionally accounting for feeder-head constraints. We first use a simplified network model with explicit security margins to determine a preliminary amount of flexibility that can be committed. The resulting schedule is then validated using a more detailed network simulation via Modelica. If network losses cause the actual response to deviate from the committed response, the electricity consumption of individual buildings is rescheduled. If the required response still cannot be achieved, the committed flexibility level is reduced. The experiments cover multiple service activation times, regulation directions, and service durations. For comparative purposes, we retain the initial optimization result as the reference estimate of the maximum potentially available flexibility, rather than interpreting the failure of a particular subsequent realization attempt as evidence that a larger flexibility level is inherently infeasible. The results demonstrated that 24.51\% of nonzero flexibility contracts were rejected due to feeder-head constraints, 67.04\% of contracts were accepted with some rescheduling and building redispatch, and only 8.09\% of contracts were accepted without any changes.

0.16

Dispersive Tree Agreements for Multi-Robot Coordination

cs.RO cs.MAmath.OC ·2026-10-22 ·Shashank A. Deshpande, Blake T. Edwards, Jonathan P. How
Abstract · 18 pages, 5 figures, 7 tables

Multi-robot coordination of nonlinear systems is naturally formulated as joint-space optimal control subject to joint constraints. Existing methods that approach joint optimality remain largely centralized and offline, while decentralized, real-time counterparts target safe navigation without pursuing the joint optimum. This paper develops a decentralized, real-time planner that approaches the joint optimum as an anytime process. We extend a recent single-robot framework that searches dispersive trees of forward-propagated commands to coordinate a multi-robot team near-optimally. We first show that the inter-robot coupling can be effectively replaced by some local constraints, under which locally grown dispersive trees contain a near-optimal agreement. It is this structure that the robots pursue by a negotiation driven by decentralized column generation. In the quiescent limit, this negotiation is shown to settle on a near-optimal agreement. Our real-time protocol, WarpWeaver, runs the negotiation asynchronously over a receding horizon and closes on quiesced agreements even when deadlines cut it short. In simulation, WarpWeaver reduces total travel time by up to 22% against centralized offline planners on the Dynobench suite and by up to 16% against a decentralized real-time baseline, and outperforms a task-specific planner on networked navigation. Hardware experiments on six robots planning onboard demonstrate real-world implementability.

0.16

LocoManiLab: A Unified Testbed for Data, Policies, Controllers, and Deployment in Humanoid Loco-Manipulation

cs.RO ·2026-10-22 ·Dewei Wang, Jinrui Han, Jiyuan Shi, Chenyun Zhang, Xianbo Lu +9 more
Abstract

Humanoid robots must coordinate locomotion and manipulation in everyday environments, yet progress is difficult to assess across differing policies, controllers, and task settings. LocoManiLab provides an integrated testbed spanning data collection and generation, policy training, evaluation, and physical deployment. Its task suite covers 234 atomic and compositional tasks across five household scene categories, supported by 51.5 hours of simulated teleoperation. The evaluation covers twelve policies with a shared controller across task and scene splits and perturbation levels, two policies across five controllers, and studies of demonstration scale and data-source mixtures. Success rates fall on unseen tasks and scenes and generally decline as visual perturbations increase, with differences in generalization and robustness across policies. For data generation, a humanoid motion generation model is fine-tuned for loco-manipulation to study the alignment of its outputs with task execution requirements and its integration into the data generation pipeline. Three simulated loco-manipulation tasks closely match real-world setups, enabling evaluation of three policies in both domains. Similar policy rankings support the reliability of simulation-based evaluation.

0.15

Behavioral Mapping of Hierarchical Dynamical Models Accelerates Bayesian Parameter Inference

eess.SY stat.MEq-bio.QM ·2026-10-22 ·Jingsong Bai, Ayush Pandey
Abstract

Bayesian parameter inference for complex nonlinear dynamical models is often limited by the cost of repeated model simulation. We present a behavioral mapping framework that uses the inferred posterior of a higher-level model to accelerate inference of a lower-level model without requiring state or parameter correspondence between them. Building on approximate simulation relations for hierarchically related models, we define the notion of behavioral compatibility and quantify the behavioral support of detailed model parameters using the reduced model posterior. We learn a level set of behavioral support using a Gaussian process and use it to filter detailed model parameter proposals before expensive simulations during Markov Chain Monte Carlo (MCMC) inference. We derive conditions under which the behavioral support level set retains all detailed model parameters that fit the experimental data within a prescribed tolerance. We apply this method to a hierarchy of gene-expression models fitted to cell-free experimental data. After initial setup costs, our approach speeds up MCMC by 8.02% for the phenomenological model and 15.21% for the mechanistic model.

0.15

FORCA: A Force-Aware Coding Agent for Contact-Rich Robot Manipulation

cs.RO ·2026-10-22 ·Xiao Tan, Yixuan Wang, Zhexi Wang, Wenzhen Yuan
Abstract · Project website: https://tanoom.github.io/forca/

Coding agents can translate language instructions into robot programs. For contact-rich manipulation, a key challenge is to generate and refine control logic that responds to physical interaction. We present FORCA, a coding agent that takes visual observations and wrist force/torque (F/T) measurements as input to write Python programs for contact probing and force-responsive motion. Between executions, the agent inspects force records alongside visual outcomes to diagnose failures and revise both motion and force-feedback control logic. FORCA requires no demonstrations, task-specific training, or simulation. Across five real-robot tasks, we demonstrate its effectiveness in three forms of contact-rich manipulation: improving manipulation accuracy, adapting to changes in object or surface properties, and regulating contact under surface variation and object motion. Compared with a vision-only variant of the same agent, FORCA improves grasp accuracy and task success and reduces contact loss in wiping and peeling.

0.14

Decentralized Bi-Level Optimization for Cooperative Orbital Transportation of a Rigid Payload

eess.SY cs.MAcs.RO ·2026-10-22 ·Sidhdharth D. Sikka, Revathy Ravuri, Hei Yin Jowett Chan, Tim Echtermeyer, Shaoshuai Mou
Abstract · 10 pages, 6 figures, 4 tables

The rapid growth of the space economy, alongside increasing orbital congestion and debris, is driving demand for reliable on-orbit transportation of large and non-cooperative objects. Performing such transportation with a swarm of small spacecraft distributes risk and removes the single point of failure inherent to monolithic architectures, but requires solving a non-convex trajectory optimization coupling every agent's cone-constrained thrust to the payload's rotational dynamics. Existing approaches span decentralized control laws with asymptotic guarantees but limited constraint enforcement, and centralized and partially centralized planners that reintroduce a single point of failure. In this work, we present a decentralized bi-level framework driven by exact gradients: the problem is split into an inner convex thrust allocation and an outer search over payload torque trajectories. This decomposition yields fuel-efficient solutions quickly, without requiring full NLP convergence, and at short timescales its performance scales with the number of agents. Theoretical convergence guarantees are established, and the method is validated through extensive Monte Carlo simulation, including closed-loop recovery from in-flight faults. These properties make the framework well suited to spacecraft swarm applications.

0.14

Beyond Binary Encoding: Qudit-Based QAOA Framework for ML Detection in $M$-QAM MIMO Systems

eess.SP ·2026-10-22 ·Chiran Siriwardena, Chanula Luckshan, Samiru Gayan, Hazer Inaltekin, Wonjae Shin
Abstract

Maximum likelihood (ML) detection in massive multiple-input multiple-output (MIMO) systems is an NP-hard combinatorial problem whose complexity grows exponentially with the number of transmit antennas and the modulation order. The Quantum Approximate Optimization Algorithm (QAOA) has recently emerged as a promising near-term quantum heuristic for such problems; however, existing qubit-based formulations require $\log_2 M$ qubits per $M$-QAM symbol, inflating both circuit width and depth. In this paper, we propose a qudit-based QAOA framework that encodes each $M$-QAM symbol directly into a pair of $d$-dimensional quantum states, where $d=\sqrt{M}$. We derive a Hermitian cost Hamiltonian via a real-imaginary decomposition of the constellation, and introduce a generalized cyclic-shift mixer that ensures ergodic exploration of the search space without binary encoding. The proposed formulation reduces the number of quantum carriers from $\mathcal{O}(N_t \log_2 M)$ to $\mathcal{O}(N_t)$ and the Hamiltonian interaction complexity from $\mathcal{O}((N_t \log_2 M)^2)$ to $\mathcal{O}(N_t^2)$, while preserving the geometric structure of the constellation. Bit-error-rate (BER) simulations for $2\times 2$ MIMO systems employing 4-QAM and 16-QAM over Rayleigh fading channels, with circuit depth $p=3$, demonstrate that the proposed detector closely approximates the performance of the optimal classical maximum-likelihood (ML) detector, while consistently outperforming the conventional minimum mean square error (MMSE) detector.

0.14

Point-Level Refinable Semantic Mapping for Spatial Understanding

cs.RO ·2026-10-22 ·Taimeng Fu, Yi Du, Zitong Zhan, Qiwei Du, Shaoshu Su +2 more
Abstract

Open-vocabulary 3D semantic maps have emerged as a key spatial memory representation for long-horizon robot autonomy, enabling robots to reason about objects, their spatial relationships, and past observations across time. However, existing object-centric semantic maps are primarily designed for static environments and rely on an append-only point integration strategy, where newly observed points are accumulated but previously integrated points are rarely reevaluated. Consequently, perception errors accumulate over time, and outdated points persist after objects move or disappear. Although recent dynamic mapping methods update object-level representations in changing scenes, they do not explicitly revise noisy or obsolete points within each object. We propose an online refinable semantic mapping framework that continuously refines object representations through probabilistic point-level confidence updating. Each object point maintains a confidence score that is updated online using multi-view temporal evidence: consistent observations reinforce reliable points, while inconsistent segmentation masks, depth measurements, and visibility cues suppress spurious and outdated points. Combined with object tracking and re-identification to preserve instance identities under motion and relocation, our framework enables object-centric semantic maps to be continuously refined rather than monotonically accumulated. Experiments demonstrate improved mapping accuracy over the previous method by 3.46% on Replica and 2.55% on ScanNet, show robustness in dynamic environments and under perception noise, and provide an effective spatial memory for object navigation and embodied question-answering with success rates of 82.4% and 76.7%.

0.14

Using Finite-Time Control to Synthesize Dynamic Backup Control Barrier Functions

eess.SY ·2026-10-22 ·Victor Turpin-Aguayo, Victor Freire, Marco M. Nicotra
Abstract

This paper presents a novel approach for the construction of control barrier functions (CBFs) for controllable linear systems subject to multiple state and input constraints. The approach is based on the Dynamic Backup CBF (DB-CBF) framework, where the control policy is parametrized using the equilibrium subspace of the system. By using a minimum-energy finite-time controller, we show that the terminal set constraint trivially reduces to having an admissible equilibrium point. The proposed DB-CBF is easy to tune (one parameter), fully takes advantage of the input range (by minimizing the control effort), and guarantees constraint enforcement (by tracking a backup equilibrium point that can be reached safely in finite time).

0.14

Adaptive Aerial Loco-Manipulation Policy Learning via LLM-Assisted Proactive Task Generation

cs.RO ·2026-10-22 ·Jingyu Ru, Xiangyuan Xie, Lixuan Jin, Jian An, Xingming Zhang +6 more
Abstract

Loco-manipulation extends the workspace of conventional manipulation by combining mobility with physical interaction, while aerial loco-manipulation further expands this capability into three-dimensional space, enabling robots to reach environments inaccessible to ground-based systems. Moving to an aerial platform, however, gives rise to a demanding control problem, since the vehicle must track trajectories accurately in three dimensions while the motion of the arm and its interaction with the environment continuously shift the center of mass, alter the inertia, and impose force and moment disturbances on the body. Conventional controllers address these effects through explicit models and estimation, whereas learning-based controllers infer dynamics from experience, yet control rewards provide no direct target for the response information retained from history, and fixed task sampling can leave persistent tracking difficulties insufficiently explored. We present ALMA, an aerial loco-manipulation framework for adaptive flight control under external arm commands. ALMA employs a contextual flight policy that adapts from recent measurements and actions, using action-conditioned prediction of future measurement representations to train its context to retain response information relevant to flight control. To expand the experience available for this adaptation, an evolutionary process uses a large language model (LLM) to analyze observed control difficulties and propose new structures of arm motion for subsequent training. In simulation, ALMA reduces position root-mean-square error (RMSE) by up to 24.7% relative to the strongest baseline, while the same policy transfers zero-shot to the aerial manipulator and reduces hover position RMSE by 79.6% during periodic arm motion relative to the real-world baseline.

0.13

ProTrav: Probabilistic Traversability with Elevation Uncertainty Propagation for Risk-aware Off-road Navigation

cs.RO ·2026-10-22 ·Chuanlin Zuo, Miloš Prágr, Arun Kumar Singh, Mohammad Loni, Reza Ghabcheloo
Abstract · 9 pages, 7 figures

Reliable traversability estimation is essential for autonomous navigation on unstructured terrain. State-of-the-art approaches typically evaluate traversability from deterministic terrain elevation, thereby overlooking spatial correlations in the terrain and failing to account for how elevation uncertainty propagates through nonlinear traversability mappings. We propose ProTrav, a probabilistic terrain representation that uses function-space sampling to generate spatially correlated terrain realizations and propagate elevation uncertainty to traversability cost. First, a Sparse Variational Gaussian Process (SVGP) reconstructs a probabilistic elevation field from observations. Then, posterior function-space sampling generates spatially correlated terrain realizations, preserving spatial dependencies between neighboring locations while propagating elevation uncertainty through geometric features into traversability cost. Consequently, a distribution of traversability cost at arbitrary location is established, where the mean and standard deviation quantify the expected traversability cost and associated uncertainty, respectively. A risk term is defined by moment-based Upper Confidence Bound (UCB) and incorporated into a Model Predictive Path Integral (MPPI) controller for risk-aware navigation. Simulation experiments demonstrate navigation success rates exceeding 90\% on different terrains, compared with below 40\% for a baseline, while reducing attitude variations. Furthermore, the proposed approach is validated on a robot traversing real-world off-road terrain.

0.13

Multimodal Progress-Residual Fusion for Anticipating Navigation Retreat in Robotic Flexible Endoscopy

cs.RO ·2026-10-22 ·Chi Kit Ng, Tianchun Wu, Jiachen Sun, Yuming Sun, Jianru Zhang +1 more
Abstract

Flexible robotic endoscopes provide access through tortuous anatomy, where friction, deformation, and wall interaction separate commanded insertion from useful distal progress. Operators interpret this mismatch and retreat to recover a navigable configuration. Recognizing the approach to that decision requires relating motor intent to distal response and the local visual scene. We introduce Multimodal Progress-Residual Fusion (MPRF), a causal model that anticipates operator retreat from commands, electromagnetic (EM) tracking, and endoscopic video. A nominal-progress model predicts the response expected in normal or successful navigation; a temporal predictor combines its EM residual with all three observed modalities. In a 150-episode colon-phantom collection, MPRF achieves 0.580 +/- 0.007 AUPRC across three seeds, compared with 0.511 for the strongest reduced-modality control. Complete multimodal histories provide the principal ranking advantage, while the residual adds an explicit intent-response representation with a small, uncertain increment over raw fusion. Analysis of frozen alarms relates warning timing to the behavioral label and explains the increase in event recall under a longer matching window. These results establish a basis for behavioral assistance during operator-controlled navigation.

0.13

LAST-V2X: Language-Driven Adaptive Scenario Synthesis and Testing for Communication-Induced Failures in V2X Safety Applications

cs.RO cs.AI ·2026-10-22 ·Keshu Wu, Hao Zhang, Rui Gan, Pei Li, Jiwan Jiang +1 more
Abstract

Vehicle-to-everything (V2X) communication enables connected vehicles to make safety-critical decisions, but delayed, missing, or incorrect information can compromise driving safety. Systematically testing such failures is challenging because traffic interactions, communication conditions, and application responses must be jointly configured to produce meaningful scenarios. This paper presents LAST-V2X, a language-driven, adaptive framework for automated scenario synthesis and safety testing of V2X applications. LAST-V2X leverages large language models (LLMs) to translate natural-language test descriptions into executable scenarios grounded in roadway geometry, V2X communication semantics, and vehicle behavior. The framework integrates physical traffic simulation with spatially aware communication modeling and message-level fault injection to evaluate how information degradation affects vehicle decisions. To establish whether an observed safety failure is caused by communication rather than the underlying traffic conditions, LAST-V2X employs matched executions under degraded and ground-truth information. Furthermore, an execution-guided refinement mechanism combines parameter search and LLM self-reflection to identify ineffective test configurations, improve scenario effectiveness, and produce minimal reproducible failure cases. Experiments across diverse CARLA environments and V2X safety applications demonstrate the framework's ability to generate reliable scenarios, distinguish communication-induced failures from unrelated safety violations, and substantially improve failure discovery through adaptive refinement. These results establish LAST-V2X as a systematic approach to language-driven, causally verified, and adaptive safety testing for connected vehicle systems.

0.13

Move, don't think: sample-efficient collision avoidance with lossy proximity sensors

cs.RO ·2026-10-22 ·Carson Kohlbrenner, Niraj Pudasaini, William Xie, Naren Sivagnanadasan, Nikolaus Correll +1 more
Abstract

Data-driven robots must learn to avoid collisions with as few demonstrations as possible to avoid collecting large datasets of destructive collision examples. We explore whole-body proximity sensors as inductive biases for learning reflexive collision-dodging behavior and find that lossy sensors with low resolution were stronger biases than high-resolution coverage. We also found that, perhaps unintuitively, anticipating when and where contact happens on the body of a robot reduced its learning sample efficiency. We investigated eight candidate representations in a reinforcement learning environment, which included latent compression, future contact anticipation, lossy corruption, and privileged state estimation. The candidate representations were tested by training over 200 humanoid robots to play a simplified version of dodgeball end-to-end under an intentionally constrained sample and compute budget.

0.12

Transfer Meta-Learning for State Space Autoencoders

eess.SY ·2026-10-22 ·Stefano De Carli, Daniele Masti, Davide Previtali, Mirko Mazzoleni
Abstract · 8 pages, 4 figures. Submitted to IFAC for possible publication

Nonlinear system identification often requires costly experimental campaigns and separate model fitting for each different instance (target system) of a given system family. To reduce this burden, meta-learning approaches reuse information acquired across related systems. In particular, in-context meta-learning enables zero-shot prediction from a brief target system context (i.e., a short input-output trajectory), but generally yields large black-box sequence predictors without exposing an explicit state space model. Conversely, compact structured state space models are suitable for simulation, estimation, and control, but are usually fitted or adapted separately for each target system. To bridge this gap, we introduce a transfer meta-learning extension of state space autoencoders. The method reuses a pretrained encoder, whose weights remain fixed, to extract a compact representation of the target system from its context. An affine parameterization maps this representation to the parameters of a compact neural state space model without target-specific adaptation; the resulting model contains orders of magnitude fewer parameters than the reference in-context predictors from the literature. We evaluate zero-shot prediction and simulation performance on held-out system instances from Buck-Boost converter and Bouc-Wen oscillator system families, comparing performance against individually fitted baseline models.

0.12

Array Rotation in Near-Field ISAC: Communication, Detection, and Estimation Under Low-Complexity Transmit Architectures

eess.SP cs.IT ·2026-10-22 ·Marziyeh Soltani, Mark F. Flanagan
Abstract

This paper studies near-field integrated sensing and communication (ISAC) with a rotatable base-station array under an exact spherical-wave propagation model. For a general transmit covariance, we derive a Cramér--Rao bound (CRB) decomposition that separates target-side geometry from transmit-covariance-dependent terms. We then specialize the framework to two representative low-complexity architectures: a single-stream superposition architecture, and a dual-waveform split architecture design. We characterize the communication-, detection-, and estimation-optimal rotations for each architecture using the exact finite-array model. In addition, we show how a second-order approximation yields compact approximate expressions for the optimal rotation angles. We also study uncertainty in the target state used for beamforming design through a correlated Gaussian perturbation model, and derive approximations of the optimal robust rotation angles for target sensing and detection. The analysis shows that the two architectures behave differently under rotation: the single-stream design can align communication and detection tasks, whereas the dual-waveform design exhibits a direct communication--detection tradeoff and can incur a larger communication loss when sensing is prioritized. Numerical results show clear gains over a fixed array orientation, especially for range estimation, and close agreement between the approximate and exact rotation choices. Detection is less affected by target-state errors than estimation in the considered cases.

0.11

Revisiting Distribution Matching for State-to-Vision Policy Distillation in Laparoscopic Robotic Manipulation

cs.RO cs.CV ·2026-10-22 ·Mingyu Sheng, Jianan Fan, Dongnan Liu, Guoyan Zheng, Ron Kikinis +1 more
Abstract · 8 pages, 3 figures, 2 tables

End-to-end control policies trained via Reinforcement Learning (RL) have emerged as an alternative to conventional surgical robotic control approaches, with recent studies exploring state-based policies trained on privileged state observations. However, such approaches rely on precise physical states that are difficult to obtain reliably in surgical environments. Vision-based RL provides an alternative by learning manipulation policies directly from RGB observations, avoiding explicit state estimation and manually engineered perception pipelines while enabling policies to acquire task-relevant visual representations. Nevertheless, obtaining vision-based policies via pure RL methods for surgical manipulation remains challenging. Thus, imitation learning (IL) becomes another solution, yet it requires a large number of expert demonstrations, which are difficult to obtain in minimally invasive surgery. In this work, we propose a framework for state-to-vision policy distillation in laparoscopic robotic manipulation. Specifically, a state-based policy with privileged observations is employed as the teacher, and a vision-based policy learns from RGB observations and robot proprioceptive information, which requires less expert demonstrations than IL and introduces superior performance than RL methods. We conduct extensive experiments on various surgical manipulation tasks in the simulated environment. Experimental results show that, in the context of laparoscopic robotic manipulation, training vision-based policies with general-purpose RL methods is challenging, whereas our proposed state-to-vision policy distillation leads to more consistent and stable performance across tasks.

0.11

DiffDreamer: Coupling World Models with Differentiable Simulation for Agile Quadrotor Control

cs.RO ·2026-10-22 ·Fanxing Li, Yuxiang Huang, Shengyang Wang, Fangyu Sun, Shuyu Wu +5 more
Abstract

Tasks such as quadrotor dynamic-gate racing require agents to anticipate external motion while maintaining precise high-frequency control. World-model-based reinforcement learning provides a natural solution by learning latent dynamics for future imagination, yet it often entangles environmental evolution and the robot's own dynamics in a single learned rollout, where model bias can affect policy optimization efficiency. We propose DiffDreamer, a differentiable reinforcement learning framework that tightly couples a learned world model with differentiable simulation for actor optimization. In one optimization step, the learned world model propagates latent states that capture external dynamics and history-dependent features, while the differentiable simulator propagates the robot's explicit physical state. We further decompose the actor objective into a critic-driven pathway for non-differentiable task rewards and a simulator-driven pathway for differentiable motion rewards. Experiments on aggressive tracking and dynamic-gate racing show that DiffDreamer improves training convergence over the model-free, model-based, and differentiable-RL baselines, and final performance over these baselines and classical planning and control methods. Additional experiments assess reward decomposition and the odometry-conditioned relative-state representation through ablations. We also provide proof-of-concept real-world flight experiments for validation.

0.10

PrefO2O: Zero-Shot Preference-Based Reinforcement Learning via Offline-to-Online Latent Preference Transfer

cs.RO cs.AI ·2026-10-22 ·Ruiqi Wang, Dezhong Zhao, Xuesu Xiao, Erdem Biyik, Byung-Cheol Min +1 more
Abstract

Preference-based reinforcement learning (PbRL) has emerged as a promising paradigm for learning robot behaviors aligned with human intent. However, its scalability is fundamentally limited by the need to collect expensive human preference labels for every new task. In this work, we present PrefO2O, an offline-to-online latent preference transfer framework that enables zero-shot PbRL on unseen target tasks without requiring any human feedback. The core idea is to learn a task-invariant latent preference space from a small amount of offline preference data on source tasks, where transferable preference structures are disentangled from task-specific variations. This learned manifold allows PrefO2O to transfer preference structures and reward priors to new tasks under distribution shift, while confidence-weighted optimal transport alignment generates reliable pseudo-preference supervision for online reward adaptation. Extensive experiments on robotic manipulation and locomotion benchmarks demonstrate that PrefO2O substantially outperforms existing preference-transfer baselines and matches or exceeds PbRL methods trained with real human feedback, while generalizing across multi-to-one and one-to-one transfer settings as well as across different robot embodiments.

0.09

IORA: Irregular Object Immobilization under Restricted Actuation

cs.RO ·2026-10-22 ·Sang-Hun Han, Changyeon Won, JunGyu Lee, Min-Gyu Park, Hae-Gon Jeon
Abstract · Preprint

Stable grasping under task-specific disturbances is important for robotic manipulation. However, most dexterous grasp synthesis methods are developed for highly actuated hands in free space, and low-DoF hands may not be able to realize the contact topologies required by these grasps. Meanwhile, support surfaces can provide useful contact forces, but are often treated only as geometric constraints. We present a support-aware grasp synthesis framework for object immobilization with low-DoF coupled hands. Given irregular object geometry, our method represents both hand and object local surface regions as graphs and searches for kinematically feasible matchings between them. We then refine the wrist pose and motor state for geometric contact feasibility and validates the refined grasps through physical rollouts. Starting from the validated grasp pose, an optional task-guided local search explores nearby grasp configurations by varying the wrist pose and motor state, and selects one according to the specified task condition. Experiments on diverse objects show that our method generates various grasp configurations that achieve high grasp success and strong disturbance resistance. Real-world experiments further demonstrate physically executable grasps which enable object immobilization in cooking scenario.

0.09

PATH: Preference-Aware Traversability Estimation for Robotic Teleoperation in Off-Road Navigation

cs.RO cs.HC ·2026-10-22 ·Xianghao Meng, Wenshuo Wang, Cheng Shao, Lewei Feng, Qingxiao Liu +2 more
Abstract

Traversability estimation is critical for ground vehicle teleoperation, yet existing methods often neglect the human operator's preferences for speed and risk tolerance. To address this gap, we propose PATH-Net, a Preference-Aware Traversability estimation Hierarchical Network that adaptively generates preference-conditioned traversability cost maps and speed profiles for off-road teleoperation. Our framework integrates visual and geometric bird's-eye-view features with short-term historical vehicle states through a multi-head shared-backbone architecture. Our PATH-Net infers traversability cost maps while optimizing speed recommendations by using maximum entropy inverse reinforcement learning and sparse speed supervision. Our method incorporates operator preference via an asymmetric feature aggregation mechanism, splitting inference into a cost branch that prioritizes high-risk traversability regions and a speed branch that outputs preference-adjusted recommended speeds. Experiments with 15 operators across 225 teleoperated trajectories show alignment with user-specific navigation strategies, achieving modified Hausdorff distance of 0.17 m (aggressive) and 0.15 m (conservative). In the questionnaire assessment, participants reported lower NASA-TLX workload scores with PATH-Net recommendations than without recommendations, decreasing from 34.7 to 22.2. These results support preference-conditioned trajectory and speed recommendation for off-road teleoperation.

0.09

People & Society 35

ABRA: Agent-Bound UI for Human-Agent Communication beyond Chat with Reusable Views

cs.HC cs.MA ·2026-10-22 ·Seon Gyeom Kim, Hannah Kim
top venuecode/data
Abstract · EMNLP 2026 System Demonstrations

Agents can create rich representations, yet these representations are rarely used as a medium for expressing user intent. We present Agent-Bound UI, an interaction model that separates reusable interaction structures from task-specific meanings. Agents bind task-specific meanings to interactions in reusable views and publish the resulting bindings as shared state, enabling humans and agents to understand and act on shared surfaces whose meanings vary across tasks. We implement ABRA, a framework-agnostic prototype of Agent-Bound UI, in which users and agents collaborate through chat and shared interactive canvases. We demonstrate ABRA through a representative use case and evaluate four LLM-backed agents on canvas authoring, owner-side consequence resolution, and cross-agent invocation. The code and demo video are available at https://github.com/megagonlabs/abra.

0.56

Generalized green and red concepts for artificial intelligence, big data, and information and communication technologies: Multidimensional framework and countermeasures

cs.CY cs.AI ·2026-10-22 ·Jinsong Wu, Yixin Xiao, Haowei Bi, Huamin Fang
published
Abstract · This was submitted February 2026 and was reviewed on hold and then approved after April 2026, the ticket number was MOD-85400

The rapid expansions of artificial intelligence (AI), big data (BD), and information and communication technologies (ICT) are driving substantial growth in electricity, water, and material demands, with data centers projected to consume 6.7--12\% of U.S. electricity by 2028 (up from 4.4\% in 2023) and AI-related water demand estimated at 4--6 billion m$^3$ annually by 2027. Existing ``green AI'' and ``red AI'' discussions have largely emphasized computational efficiency and carbon emissions, providing limited guidance for lifecycle trade-offs, service-versus-use impacts, and broader sustainability externalities beyond energy. To bridge this gap, this paper proposes generalized green AI (GGAI) and generalized red AI (GRAI) and extends the taxonomy to red BD (RBD) and red ICT (RICT) as well as red electronic and electrical engineering (REEE) \& red digital technologies (RDT) under a unified Environmental-Economic-Social-Technical/Operational (E/Ec/S/T) framework. Based on representative footprint indicators in the large-model era (electricity, carbon, water, and e-waste), we analyze how risks propagate across AI-BD-ICT stacks and across service-use boundaries. We further organize actionable countermeasures into a four-tier taxonomy aligned with the Environmental-Economic-Social-Technical/Operational framework: metric-level reform for accountability, revision of information and communication technologies design principles and demand shaping, efficiency-oriented LLM development practices (data, model, training, and scaling), and energy-infrastructure transition. The proposed framework operationalizes the Environmental-Economic-Social-Technical/Operational-aligned assessment and intervention prioritization across the AI-BD-ICT stacks, enabling sustainability-by-design without sacrificing utility.

0.49

Impacts of generative AI skill development on academic adaptation and learning outcomes among university students

cs.HC cs.AI ·2026-10-22 ·Lei Fan, Fangxue Liu
Abstract

Understanding how university students perceive the development of their generative artificial intelligence (GenAI) skills over time can clarify how GenAI use relates to learning and academic outcomes. This study examined perceived skill-development trajectories, changes in learning practices, and associated academic gains using a retrospective questionnaire with 313 analyzed responses and interviews with 18 students across subject areas. Descriptive and comparative analyses and exploratory structural equation modeling (SEM) were integrated with thematic analysis. Perceived developmental trajectories were heterogeneous: 34.2% of students reported rapid early improvement followed by a plateau, 28.4% steady improvement, and 21.1% delayed acceleration. Development was also uneven across skills, with improvement in evaluating GenAI outputs perceived as more challenging; assignments were the most frequently selected driver of skill growth. Interviews further indicated increasingly specific prompting, tailored verification, and selective delegation, although some students reported stable routines, dependence, skipped reasoning, or improvements attributable partly to better GenAI tools. Academic gains were reported more often for assignments than for examinations. The SEM indicated a positive indirect association between perceived skill growth and academic gains through adaptation of assignment processes, whereas academic gains did not clearly differ across developmental trajectory groups after multiplicity correction. The findings suggest that perceived GenAI skill development varies in both trajectory and skill domain and is associated with changes in learning practices and differentiated academic benefits.

0.47

Metacognitive Scaffolds Create Productive Friction for AI-Assisted Knowledge Work

cs.HC cs.AI ·2026-10-22 ·Viktor Kewenig, Lev Tankelevitch, Sean Rintel, Advait Sarkar, Abigail Sellen
Abstract · 46 pages, 11 figures, 17 tables

Generative AI imposes new metacognitive demands on knowledge workers as their work shifts from production to planning, evaluation, and orchestration. In a randomized, between-subjects experiment on AI-assisted writing (297 workers, 594 essays), we test dynamically-generated metacognitive scaffolds supporting goal specification and task decomposition. We find that scaffolds improve writing quality, with this boost persisting in short-lag transfer when scaffolds are removed. We also find evidence of improved topic recall three days later. These effects are consistent with observed changes in how people learn to prompt AI systems after exposure to scaffolding. However, scaffolded participants rate the AI as less helpful and more cognitively demanding even as they perform better, suggesting that subjective ease can understate the value of support. Our findings make the case for metacognitive scaffolds in AI-assisted knowledge work, revealing a productive friction worth designing for as workers increasingly delegate work to agentic AI.

0.46

Big World, Little Me: Comparing a First-Person VR Walkthrough Video and Static Slides for Communicating Height-Related Accessibility Barriers

cs.HC ·2026-10-22 ·Chan In Sio, Alex Mann, Qian Ming Jia, Xuecheng Guo, Yan Zhuo +3 more
published
Abstract · 14 pages, 2 figures. Accepted for presentation and publication at ICVRT 2026. Author manuscript

People with dwarfism and other forms of short stature may encounter barriers created by fixed-height environments and discriminatory social assumptions. This paper evaluates Big World, Little Me, a 10 min 52 s first-person walkthrough recorded from a Unity-based VR environment. The narrative depicts domestic reachability, public-transport, and employment barriers. In a non-randomised between-subjects study, 100 participants aged 15--48 years viewed either the fixed-duration audiovisual walkthrough or a silent, self-paced 27-slide presentation representing the same principal events. Narrative engagement was higher for the walkthrough ($M=5.54$, $SD=0.51$) than for the slides ($M=4.24$, $SD=0.73$), mean difference $=1.30$, 95% CI [1.05, 1.55], Hedges' $g=2.05$, $p<.001$. State empathy was also higher for the walkthrough ($M=3.15$, $SD=0.33$) than for the slides ($M=2.21$, $SD=0.60$), mean difference $=0.94$, 95% CI [0.74, 1.13], Hedges' $g=1.91$, $p<.001$. Exploratory supportive-intention ratings increased more in the walkthrough condition than in the slide condition. Participants also rated the walkthrough as more dynamic, continuous, spatially clear, and information-rich. The findings indicate stronger immediate responses to the complete walkthrough format, but do not isolate the effects of motion, continuity, pacing, information quantity, or participant control. The walkthrough should be understood as screen-based VR-authored media rather than immersive VR or a reproduction of lived experience.

0.45

Abstract Priorities, Concrete Rewards: How Stated Preferences Relate to What People Reward in AI Responses

cs.HC cs.AIcs.CL ·2026-10-22 ·Zihan Chen, Lei Nico Zheng, Di Zhu
code/data
Abstract · 27 pages, 5 figures, 30 tables. Code and data are available from the authors upon request

AI assistants built on large language models (LLMs) are trained on human feedback and increasingly personalized to users' stated preferences. Yet how closely these preferences reflect what people reward in responses remains unclear. We compare stated and revealed preferences in the public PRISM dataset, in which 1,396 participants reported how much seven response attributes mattered to them and then rated 29,347 responses from 21 LLMs. Among responses to the same prompt, features the survey never asked about, such as response length, formatting and refusal language, were at least as strongly associated with ratings as text proxies of the surveyed attributes. Whether stated priorities matched ratings depended on how attributes were measured: participants who prioritized an attribute favoured responses from models that others perceived as strong on it, but not responses that scored higher on text proxies for it. Written feedback was more telling: participants who asked for shorter answers, even in earlier conversations, gave length less weight in their ratings, and those asking for longer answers gave it more. Their relative priorities over the surveyed attributes, however, did not predict this weight. Reasons given for choices reflected perceived attributes more than measured features. Participants also differed stably in how much they weighted several concrete features, especially length, but with a median of 6 conversations per participant, these weights were estimated poorly. The largest gain from personal information in predicting held-out choices was 0.13 percentage points, compared with 1.11 for a stronger pooled model. Stated priorities thus track perceived attributes of AI responses but say little about the concrete features that ratings reward, a distinction that matters for preference elicitation and AI personalization.

0.42

The Capability Transfer: Cohort Reversal, Codified Expertise, and the Adaptability Tax of AI Transformation

cs.CY econ.GN ·2026-10-22 ·Ina Catrinescu
Abstract · 29 pages, 6 tables, 6 figures. Working paper. Also available at SSRN: https://ssrn.com/abstract=7445098. JEL: J24, O33, M54

Most analyses of AI and employment ask which jobs are exposed. This paper asks what happens to the pipeline that produces experts. It extends the model of Caosun and Aral (2026) from a single worker to a firm with two cohorts, juniors and seniors, and studies how expertise formation changes when codified work is automated. Because junior work is the most codified, demand for juniors falls first; the stock of seniors falls later but further, as the pipeline stops replacing the experts who retire. In the baseline calibration, junior demand bottoms out about a fifth below its 2022 level before recovering most of that ground, while the senior stock falls by a quarter and later halves. Because expertise forms through unassisted practice on work that AI cannot do, the time needed to develop an expert from within rises from about five years to nine. The costs fall on a different party from the one that decides on adoption: a manager with a three-year horizon sets AI use to maximise near-term output, cutting the wage bill by 21 percent while the capacity of the firm on work AI cannot do falls by two-fifths and its output on that work goes unserved once the pipeline runs thin. Two key parameters are not yet measured, so the results are scenarios rather than forecasts. The paper proposes three indicators organisations can track: the unaided practice juniors get on work AI cannot do, senior time spent on such work, and the years it takes to develop an expert.

0.35

AusClinBench: a deterministically scored benchmark of large language models against current Australian clinical policy

cs.CY cs.CL ·2026-10-22 ·Daniel S. Epstein
benchmarkhot direction
Abstract · 19 pages, 3 figures. Technical report, version 1.4.1 of the benchmark. Not peer reviewed. Answer keys not yet independently reviewed. Result…

AI models already inform clinical decisions and draft notes in Australian clinical software, usually without the practitioner being told which model is used. Existing medical benchmarks are written for other jurisdictions. We wrote 142 questions from Australian guidelines and schedules across five domains (PBS authority, MBS billing, Ahpra obligations, triage red flags, immunisation and screening), each with one verifiable answer scored by exact match. Each question ran blind and with the governing clause supplied, which separates failure to recall the current rule from failure to apply it. We tested 28 models between July and October 2026. Blind accuracy ranged from 51.5% to 92.7%, and the highest-scoring models could not be separated. Of 791 failed model-question pairs, 684 (86%) were wrong blind and right with the clause. Confident-error rates ranged from 6.6% to 47.8%. On five immunisation questions where the guideline gives no directive, 82% of answers stated one. On 20 Aboriginal and Torres Strait Islander health questions, models did no worse overall, but on a few the wrong answers gave the rule for the general population, an older rule or another country's rule. We propose treating the model as an ingredient of a clinical product, with a minimum standard on local rules that sits beside product evaluation. The keys are one general practitioner's and are not yet independently reviewed. The benchmark measures base models, not clinical products, and does not license any model for clinical use.

0.35

Multi-Material Topology Optimization of Phononic Crystals for Targeted Elastic Wave Bandgap

cs.CE ·2026-10-22 ·Bir Bahadur Chaudhary, Wenfeng Li, Ting Chen, Xinmai Yang, Candan Tamerler +2 more
Abstract

Phononic crystals are periodic elastic structures that exhibit frequency bandgaps within which elastic waves cannot propagate. The formation and width of these bandgaps depend on the spatial distribution of material properties within periodic phononic unit cells. In this work, we propose novel topology optimization formulations to achieve multi-material design of phononic crystals for bandgap optimization at targeted frequencies, where we discover the influence of spatially varying properties of base materials on the band structures that block and permit elastic wave propagation. Specifically, we model wave propagation by Floquet-Bloch theory and solve the resulting eigenvalue problem by the finite element method along the boundary of the irreducible Brillouin zone. We first propose a multivariate weight-based method via a novel interpolation scheme that maps the design variables of an arbitrary number of candidate materials into effective density and stiffness for elastic wave-propagation calculations, where bandgaps arise from the contrast in the stiffness and inertia among material constituents. We then develop two formulations for bandgap optimization. The first formulation maximizes the gap between two adjacent frequency bands using a bound-based approach, where we introduce two auxiliary scalar variables as additional design variables and incorporate them directly into the design objective, yielding a linear, differentiable design objective, with the eigenfrequencies entering the formulation through associated constraints. The second formulation places a bandgap over a user-defined target frequency range through a dynamic band-pairing strategy, where we adaptively identify the relevant pair of adjacent bands amid optimization iterations. Towards the end, we validate the bandgap performance of our designs against the transmission loss spectra computed from COMSOL Multiphysics.

0.35

Working Through Representation: Malleable Workspaces for Human-Agent Knowledge Work

cs.HC ·2026-10-22 ·Nan Chen, Long Ling, Luna K. Qiu, Zilong Wang, Yuqing Yang
hot direction
Abstract

Generative AI supports knowledge work, yet conversational histories may not capture the evolving organization that guides subsequent action. We introduce the Malleable Workspace, a conceptual model of how humans and agents form and revise a Work Representation and use its organization to advance work within a wider Work Environment. We developed it iteratively through conceptual analysis, comparison of 58 papers, and the construction and three-month longitudinal study of MalleableSpace with seven experienced agent users. Systems varied in what they represented as units and relations. System-assembled context was common; more systems supported local incorporation of agent outputs than reorganization, and fewer supported consequential agent-driven changes beyond the representation. The field study showed how representational roles and organization developed in use. These observations refined the model's account of how representation-grounded action depends on and can change the Work Environment. We contribute a conceptual and analytic foundation for representation-centered human--agent design.

0.31

Do We Need to "Be There" to Say Goodbye? Comparing Participatory Agency and Spatial Presence Across Digital Farewell Formats

cs.HC ·2026-10-22 ·Chan In Sio, Alex Mann, Qian Ming Jia, Lik-Hang Lee
published
Abstract · 10 pages, 4 figures. Accepted for publication in IEEE Transactions on Visualization and Computer Graphics (ISMAR 2026). Author accepted manu…

Digitally mediated attachment is increasingly common, yet the mechanisms of symbolic closure in hybrid physical--digital relationships remain underexplored. We address an XR design question: how different mediated farewell formats relate to perceived symbolic closure, spatial presence, and bereavement-related emotional intensity. Using "cloud petting" as a controlled proxy of digitally mediated attachment, we compare three farewell formats: (1) lightweight social messaging, (2) a revisitable web memorial, and (3) immersive VR ritual participation. These formats differ in symbolic participation, spatial presence, and sensorimotor ritual enactment. In a mixed-methods study ($N=45$), Website and VR showed significantly higher perceived symbolic closure than Messaging, while the present sample did not detect a significant closure advantage for VR over Website. VR was also associated with significantly stronger grief-related responses on the Pet Bereavement Questionnaire, whereas negative-affect change measured by the Positive and Negative Affect Schedule did not differ significantly across modalities. This pattern motivates a preliminary, exploratory, mechanism-level Agency-Amplification account: formats supporting structured farewell participation were associated with higher perceived closure, whereas the VR format, combining technical immersion, spatial presence, and sensorimotor ritual enactment, was associated with stronger loss-oriented emotional processing. Because the formats also bundled differences in participation, revisitation, technical immersion, and spatial presence, the study provides preliminary design guidance without isolating the causal contribution of any single feature.

0.31

Real-to-Sim Calibration of Bosch Etching Simulators with Few-Shot XSEM Data and Hybrid Search

cs.CE cs.LG ·2026-10-22 ·Xinyu Jia, Jainil Shah, Rahul Tripathi, Vijay Gupta, Zhihong Chen
Abstract · 38 pages, 12 figures, 8 tables

The Bosch process is widely used in MEMS and TSV fabrication, but accurate prediction of etching outcomes remains difficult without calibration against real measurements, which are sparse and costly to obtain. This work presents a few-shot real-to-sim calibration framework that decouples simulator parameters from direct process variables to infer a recipe-specific simulator parameter vector from sparse measurements, enabling reliable prediction of etch depth and sidewall angle under held-out process conditions. We first formulate Bosch profile prediction as a few-shot real-to-sim calibration problem, where a small set of real XSEM measurements is used to identify a recipe-specific ViennaPS parameter vector $z_r^*$ that minimizes the discrepancy between simulated and measured etch depth and sidewall angle. To search this parameter space efficiently, the framework combines local exploration, LLM-guided candidate generation, and XGBoost-based surrogate screening, while retaining ViennaPS simulation as the final verification step for each selected candidate. Experiments on two Bosch etching recipes show that the proposed framework accurately predicts held-out etching results, achieving normalized global errors of 0.340 and 0.249 for the 2 recipes respectively.

0.30

Structure-preserving reduced-order models of thin panels in supersonic and hypersonic flow

cs.CE cs.LGmath.NA ·2026-10-22 ·Miguel A. Aguilo
Abstract · 68 pages, 15 figures

We build structure-preserving reduced-order models for both challenge problems of the AePW-4 High-Speed Working Group. They are a clamped thin panel at Mach 1.92 (RC-19) and a cantilever plate at Mach 5.85 (HyMAX). The published data for each case hold no time-resolved response and no repeated runs. The physics must therefore be carried by the form of the model. One construction gives two models per problem. A Ritz-Galerkin reference model describes the plate. A learned model keeps the reference model's exact potential, fits the remaining generalized forces, and trains only a dissipation that cannot add energy. Its energy balance holds for any weights by construction. On the trained model the worst relative residual is 2.2e-15. On RC-19, shock relations predict a pressure ratio of 1.521 across the panel. Pressure-sensitive paint measured the same value. Thirty runs of the learned model place the mean dominant frequency within 1.0% of the measured 236 Hz. At one periodic and one chaotic condition, the learned model reproduces the response regime of a published reduced-order model in all sixteen runs. It does so without that model's 113 Hz spectral peak, which the measurements do not show. Measured displacements can also correct the chaotic response through one more force that respects the energy balance. We tested this on synthetic data at the experiment's 5 kHz sampling rate. The state error is 1.25% of the panel thickness without noise and 3.5% with 10 micrometers of noise. On HyMAX the learned model follows the reference model's trailing-edge displacement for 65 ms with a relative error of 3.7%. These results show that a model whose structure carries the physics can match measured flutter behavior when the available data cannot train it.

0.30

Multi-Material Topology Optimization for Targeted Structural Dynamic Vibration Attenuation

cs.CE ·2026-10-22 ·Bir Bahadur Chaudhary, Bidhan Pandey, Bin Ouyang, Shiguang Deng
Abstract

Resonance-induced excessive vibrations can degrade performance, reduce reliability, and shorten the service life of engineering structures. To reduce the risk of resonance, structural natural frequencies are generally engineered to be shifted away from excitation frequencies. To achieve these frequency shifts, topology optimization provides a systematic approach to tailor structural stiffness and mass as an effective inverse design approach. Multi-material topology optimization exploits the high-dimensional design space of spatially varying material constituents by simultaneously optimizing material selection and spatial distribution over the design domain, offering great flexibility in controlling natural frequencies. This study proposes a multi-material topology optimization that attenuates excessive structural dynamic vibrations in free-vibration scenarios via a novel multivariate weight-based interpolation method. Additionally, to address non-smoothness due to mode switching amid optimization, we employ a bound formulation with an auxiliary scalar design variable to replace the max-min eigenvalue objective with constraints on individual eigenvalues. Furthermore, we study and compare three optimization formulations for multi-material design to attenuate vibrations, including maximizing fundamental eigenfrequency, maximizing the gap between adjacent eigenfrequencies, and excluding frequencies from targeted frequency bands, which are smoothly approximated by a proposed rational polynomial formulation. Numerical examples are presented to demonstrate the effectiveness and applicability of our proposed method.

0.30

ClinGuard: a governance framework for reliable, transparent, and auditable clinical coordination

cs.CY ·2026-10-22 ·Kaiyuan Wu, Aditya Nagori, Nicoleta J. Economou, Robert M. Califf, Rishikesan Kamaleswaran
published
Abstract · Perspective, accepted at npj Health Systems

ClinGuard is a model-agnostic governance framework that treats clinical coordination as patient-scoped, policy-bound transactions rather than free text. An independent sentinel layer validates and governs agent-proposed actions, escalates when needed, and executes approved actions through limited-privilege connectors. Each transaction is linked to a tamper-evident audit record, while an outcomes monitor informs future autonomy decisions. Using MIMIC-IV-FHIR and skilled-nursing-facility placement, we distinguish retrospective demonstration from proposed prospective evaluation in this work.

0.28

GELLI: A Generator for Experiential Language Learning Interactions

cs.HC ·2026-10-22 ·Danilo Symonette, Alan Y. Cheng, Taiga Ikedo, Jose Manuel Rosales Juarez, Roy Pea +3 more
Abstract

Learning a second language requires extensive practice across situations personal to each learner, yet the most effective existing approaches --- narrative situations designed around specific learning goals and delivered through media --- demand substantial authoring effort, limiting practice to the scope of what was built. We present GELLI, a system that generates narrative practice worlds from a learner's goals: situated scenes with characters to converse with, support alongside the interaction, and reflection, all produced at runtime with no pre authored content. In a within-subjects study with 40 learners of Japanese, GELLI produced greater vocabulary recall and confidence than expert-authored videos targeting the same goals, with no significant difference in speaking performance. Learners turned remembered or imagined situations into experiences they could shape and learn through. Support made experimentation feel safe, while unexpected narrative developments and language suggestions gave them new possibilities to pursue beyond what they initially intended to practice.

0.28

Making Generative AI Workable at Work: Accountability, Localization, and Representation in Kenyan and Nigerian Small Businesses

cs.HC cs.CY ·2026-10-22 ·Najeeb Gambo Abdulhamid, Stephanie Nyairo, Liz Ankrah, Jacki O'Neill
position
Abstract

Generative AI is widely framed as democratizing skilled work by lowering the barrier to producing professional output. Drawing on 21 semi-structured interviews with 27 workers across 21 small and medium-sized businesses (SMBs) in Kenya and Nigeria (July to October 2023), we find that availability alone did not make participation workable. In participants' accounts, drafting was rarely the hard part. The consequential work was making AI-assisted drafts defensible enough to circulate to clients. We identify five intertwined practices through which participants made generative AI workable: using it as draft labour, steering outputs through iterative interaction, switching and comparing across tools, performing release work to make outputs accountable before circulation, and repairing context misfit when outputs failed local requirements. We make two central contributions. First, we specify release work as a distinct, boundary-facing form of accountability labour, evidenced through verification gates, traceability reconstruction when sources fail, and non-delegation boundaries that decide what can ship. Second, we show how localization and representational failures, which we call context misfit, operate in everyday work as a recurring breakdown-and-repair class, including failures of local language, institutional knowledge, and the credible representation of African people and places, and we document the repair repertoire they demanded. Reframing democratization as workable participation, we argue that reported gains were conditional on this accountability and repair labour, and we translate the findings into four design directions for workplace generative AI. We treat Kenya and Nigeria as two situated cases, not as a representative sample of SMBs or the Global South.

0.28

What the submission does not show: Complexifying cognitive offloading in students' mathematical work with LLMs

cs.CY math.HO ·2026-10-22 ·Igor' Kontorovich, Miriam N. Wallach
Abstract · 44 pages. Preprint under consideration at Digital Experiences in Mathematics Education; not peer reviewed

A common concern is that using large language models (LLMs) lets students offload mathematical work that tasks are designed to elicit. Yet such claims often rest on outcomes (e.g., submitted solutions), rather than on evidence of students' work with LLMs. We followed 23 first-year linear algebra students as they solved three differently designed homework tasks with LLMs. Drawing on self-regulated learning and choice-affluent environments, we interpreted each of 429 prompts as a trace of a regulatory event and compared 59 chatlogs with the problem solutions the students submitted. Prompts formed five structures, casting the LLM mainly as a solver, explainer, or validator. The mix of structures shifted across the tasks, and ten of the 15 students who completed all three tasks changed their dominant structure. Most final submissions stayed close to LLM output, including those that followed exchanges in which students had explained, conjectured, and checked their own ideas. We conclude that an LLM-aided submission is not a trustworthy marker of students' offloading of the mathematical work, and that offloading needs to be examined in each regulatory space and part by part rather than as a whole.

0.28

What Does AI Success Look Like? A Mixed-Methods Analysis of Supranational AI Governance Instruments

cs.CY ·2026-10-22 ·Delaram Golpayegani, Marta Lasek-Markey, Monique Munarini, Arjumand Younus, Aphra Kerr +1 more
Abstract

In the last decade and in particular since 2018, supranational institutions have published policies, guidelines, and most recently regulations to govern development and use of Artificial Intelligence (AI). This paper aims to investigate the themes and dimensions of ``AI success'' set forth by instruments of AI governance issued by supranational bodies. To this end, we use a mixed-methods approach, combining manual thematic analysis and automated topic modelling approaches, to analyse a curated corpus of 13 documents issued by supranational bodies, including the European Union bodies, OECD, G20, UNESCO, United Nations, and the Council of Europe between 2018-2025. Based on the analysis and through the lens of sociology of expectations, we propose a novel categorisation scheme for AI success themes emerging from supranational bodies' documents. Drawing on the findings, we advance a set of policy recommendations, setting forward the need to move beyond high-level principles toward more coordinated approaches to AI governance.

0.28

Hyperreduction accelerates finite element model updating for the discovery of neural materials

cs.CE cs.AI ·2026-10-22 ·Hagen Holthusen, Matthias Knipper, Kevin Linka
Abstract · 42 pages, 23 figures, 1 table

Finite element model updating is among the most general approaches to discover constitutive models from structural experiments: since the displacement field is computed rather than measured, it accommodates arbitrary geometries, boundary conditions, and sparse or noisy data. However, every loss evaluation requires a nonlinear finite element solve per load step, and training neural materials requires many such evaluations. Intrusive model order reduction lowers this cost while retaining boundary conditions and variational structure. How can a reduced model remain accurate while the network weights change at every training step? We combine proper orthogonal decomposition and energy-conserving sampling and weighting with adjoint sensitivities of the reduced model, and adapt both the trial space and the sampled elements along the optimization path. The constitutive law is a polyconvex neural network: a logarithmic normalization term, together with a zero-anchored input convex architecture and a growth term, renders the energy polyconvex, coercive, and non-negative by construction, independent of the trained weights. We demonstrate the approach on two experimental datasets and one synthetic three-dimensional example. For a two-phase specimen, training time decreases from nine hours to 41 minutes compared with a full-order identification, without loss of accuracy. In the three-dimensional example, six modes replace 48 672 degrees of freedom, 116 of 15 625 elements suffice to evaluate the internal forces, and training takes 13 minutes. In all cases, the identified material reproduces the measured reaction force and predicts the displacement field from the force alone. Hyperreduction makes the discovery of neural materials from structural data affordable while retaining the advantages of finite element model updating.

0.28

DLT Terms 100: a learner's glossary based on ISO 22739:2024

cs.CY cs.CR ·2026-10-22 ·Yusuke Kaneko, Takumi Aita, Kohei Ichihara, Hiroki Watanabe
unusual
Abstract

We propose ``DLT Terms 100'' aimed at improving literacy regarding blockchain (distributed ledger technology: DLT). The pace of DLT's social implementation has been slower than initially expected, partly due to a shortage of engineers and literate individuals. In particular, the discrepancy between business and technical terminology, and the confusion caused by the repurposing of existing financial terms, are barriers to understanding the concepts and technologies. Our proposed glossary is based on ISO 22739:2024, and categorizes and labels terms based on their frequency of appearance in representative domestic and international literature and their difficulty level.

0.27

Navigating human-robot interaction challenges in private and residential aged care homes

cs.HC cs.CVcs.RO ·2026-10-22 ·David Ahmedt-Aristizabal, Mohammad Ali Armin, Akila Pemasiri, Harshala Gammulle, Tharindu Fernando +2 more
published
Abstract · Author-accepted manuscript of a book chapter published online in Elsevier's Reference Module in Materials Science and Materials Engineering …

This chapter explores how robotic systems and ambient intelligence can enhance care for older adults in residential aged care (RAC) homes and private homes. Although these settings differ, we frame both as extreme and challenging environments for human-robot interaction (HRI), characterised by cluttered and dynamic layouts, diverse resident needs, and limited on-site technical support. We review embodied and non-embodied systems that assist with daily living, safety monitoring, and social engagement, highlighting hybrid deployments where robots complement fixed and wearable sensors. To ensure these technologies are usable and effective, design considerations consider the cognitive, physical, and social needs of older adults. Interaction design prioritises accessibility for older adults with low digital literacy, using robust but simple voice interfaces, small sets of non-verbal cues, accessible screen-based companions, and safe object handovers. Privacy, trust, and acceptability are addressed via practical strategies such as on-device inference, data minimisation, visible consent controls, and transparent data handling aligned with emerging artificial intelligence (AI) and privacy regulation. Evaluation emphasises real-world metrics including false alarms per day, time to recover from faults, staff workload, and long-term system reliability. We close by outlining open research challenges and future directions for scalable, regulation-ready deployments that support aging with dignity, safety, and independence.

0.27

How Different, or Different How? Stance Granularity in Viewpoint Selection for Dialog Systems that Engage Users with Alternative Views

cs.HC cs.CL ·2026-10-22 ·Morita Tarvirdians, Reeve Lorena, Catharine Oertel
Abstract

Systems that aim to help a user engage with viewpoints other than their own have to decide how different those viewpoints should be. Viewpoints are mainly represented on a scalar scale, too coarse to distinguish viewpoints that disagree with the user about different things. We introduce a selection policy in which a bandit sets how far viewpoints may sit from the user, and the policy selects the most novel viewpoint within that bound. We vary only its stance encoder, comparing a scalar score with Perspectivized Stance Vectors, which break a position into separate stances on concepts. We integrate it inside a dialog system and run a between-subjects study (N=63). Concept-level stance produces more varied exposure sequences and broader coverage of the topic's positions, but does not change how deeply users justify viewpoints or how accurately they reconstruct them. Finally, we analyze user--viewpoint relations at the concept level, showing that users reconstruct a viewpoint least accurately on concepts that one side took no position on, rather than on concepts where the two disagreed. Our findings open up opportunities for dialog systems that support engagement with alternative viewpoints by acting on which concepts a user and a viewpoint disagree on, rather than on distance alone.

0.26

An AI Factory for Technology Education: An Agentic Pipeline with Closed Loop Audience Feedback

cs.CY ·2026-10-22 ·Josue Hernandez, Hector R. Amado-Salvatierra, Rocael Hernandez Rizzardini
Abstract · 20 pages, 3 figures

Technology tools used for professional upskilling change on a timescale of weeks, and written guidance about them ages at the same pace. Large language models have made it inexpensive to draft a tutorial or a course outline, but most deployed generation systems still produce a piece of content once and never check it against what an audience actually does with it. This paper describes an agentic pipeline, organized as a typed workflow graph, that treats instructional content production as a manufacturing process with explicit checkpoints rather than a single generation call. Four node classes structure the graph: deterministic scripts, model steps bound to a fixed rubric, open-ended agent phases, and human approval gates, arranged along a forward production path and closed by a feedback cycle that rewrites generation rules from post level engagement telemetry. Production of content is split into two layers: a demand layer, built from video and community sources, that shapes topic selection and register, and an authority layer, built from official documentation, that every published claim must trace back to. Over five months of continuous operation, the deployed system published 26 documented guides at roughly 1.5 hours of operator time per week, and the feedback loop identified four measurable engagement effects that became binding rules for later runs, including a hook style that outperformed baseline by a factor of 1.83 in its first analysis window.

0.25

Supporting or Replacing Students' Reasoning? Cognitive Offloading in AI-Assisted Proof Learning

cs.HC ·2026-10-22 ·Eason Chen, Sophia Judicke, Kayla Beigh, Xinyi Tang, Isabel Wang +16 more
Abstract · 14 pages. This manuscript presents a substantially revised and extended analysis of the classroom study reported in arXiv:2602.18807 and the…

We studied 153 students in an undergraduate proof-based mathematics course with staggered-access to an AI support system. Students were randomly assigned to receive an AI chatbot and a homework-feedback interface earlier or later in the course. During the period when only the early-access group had access, their homework scores were 2.56 points higher, but exam performance did not differ significantly between groups, suggesting that improved AI-supported homework performance did not translate into detectable gains on a subsequent exam. After all students received access, every additional 100 messages sent to the AI chatbot was associated with a 3.02-point lower score on the subsequent exam. A mediation analysis showed a significant indirect association linking lower baseline self-efficacy to greater chatbot use and, in turn, lower exam scores: a one-point decrease in self-efficacy corresponded to 0.62 fewer exam points through chatbot use. These findings raise concerns that students may use the chatbot to offload their cognitive reasoning. Use of the homework-feedback interface, which required students to submit their proof attempts before receiving feedback, was not negatively associated with exam performance. We recommend against using a chatbot for AI tutoring and suggest exploring a tutor embedded in the course activities.

0.25

In-Browser Human-AI Co-Writing for Counterspeech: An Exploratory User Study of Agency, Perceived Response Quality, and Time

cs.SI cs.HC ·2026-10-22 ·Tammy Schmidt, Luise Koch
Abstract · Pre-Publishing for the 10th Workshop on Online Abuse and Harm 2026

Writing counterspeech that challenges online hate speech can be cognitively and emotionally demanding, whereas fully automated generation may weaken users' sense of voice and responsibility. We present an in-browser AI-assisted co-writing tool that enables users to contribute their own ideas and offers three editable suggestions generated by a large language model using dynamically retrieved hate speech-counterspeech examples from the Multitarget CONAN dataset. We evaluated the tool in an exploratory within-subject study with ten participants, complemented by think-aloud observations and semi-structured interviews with half of the participants. Participants wrote responses to three self-selected hate speech instances without assistance and then repeated the task with the tool. Final responses were subsequently peer-rated for empathy, non-toxicity, relevance, specificity, and persuasiveness. Participants evaluated the interface positively, and qualitative accounts linked user input, selection among alternatives, and small edits to perceived agency and responsibility. Assisted responses received higher descriptive mean ratings across all five proxies, with the largest difference for empathy. Mean composition time increased from 2.52 to 3.88 minutes. Qualitative findings further revealed a tension between empathy-oriented outputs and users' desire to express legitimate anger or direct disagreement. We contribute exploratory design guidance for treating AI-assisted counterspeech as deliberative co-writing rather than automated text production.

0.25

A Topology-Aware Quantum-Classical Framework for Multi-Target Active Debris Removal using Gaussian Boson Sampling

cs.CE cs.ETquant-ph ·2026-10-22 ·Komal Pandya, Saiyam Sakhuja, Nagaraj Kattaian
benchmark
Abstract

The continued growth of orbital debris threatens the safety and sustainability of near-Earth space, making Active Debris Removal (ADR) an important mitigation strategy. Selecting $k$ targets from $N$ candidates yields a combinatorial search space of size $\binom{N}{k}$, making joint target selection and mission sequencing computationally demanding. We present a topology-aware quantum-classical framework integrating orbital mechanics, weighted orbital-transfer graphs, Gaussian Boson Sampling (GBS), and multi-objective evolutionary optimization. Seven target-prioritization strategies represent physical consequence, orbital persistence, conjunction risk, network centrality, community structure, inferred source families, and random selection. GBS generates candidate target subsets from the weighted graph, which are then evaluated using PyGMO-based evolutionary optimization to identify servicing sequences that jointly minimize total velocity increment ($Δv$) and mission duration, accounting for modeled timing, propulsion, and mass feasibility. We evaluate the framework across three benchmark environments, SOCRATES Minefield, Collision Cloud, and Random, with candidate-pool sizes $N\in\{15,18,22,25\}$ and target cardinalities $k\in\{3,5\}$. The experiments characterize how benchmark topology, prioritization strategy, and problem scale relate to sampling behavior and downstream mission performance. By separating candidate-subset generation from detailed trajectory optimization, the framework offers a topology-aware approach to focusing computational effort on a limited set of candidate ADR missions.

0.24

AI-Native Academia Should Build a Community Verification Layer

cs.CY ·2026-10-22 ·Haotian Zong, Luhao Zhang
Abstract

This paper argues that AI-native academia should build a Community Verification Layer: interoperable, policy-governed infrastructure that converts researchers' routine reading and checking into persistent, claim- and version-linked evidence, while routing scarce deeper verification toward uncertain, consequential, and neglected research. AI is lowering the cost of producing plausible research artifacts faster than the cost of accountable evaluation, yet much of the evaluation researchers already perform while reading, citing, implementing, or extending work is never made reusable. We propose a tiered ladder from comprehension notes to claim checks, audits, artifact execution, reproduction, replication, and downstream validation. The layer would elicit judgments before revealing social signals, preserve multidimensional disagreement, privately verify contributors while allowing flexible public disclosure, allocate attention under expertise and coverage constraints, and make verification labor citable, with credit tied to the quality and depth of the evidence. We distinguish this design from prior open and post-publication review and outline governance requirements and potential counterarguments.

0.24

Your Home Remembers: A Voice-Enabled Household Memory Assistant for Conversational Retrieval

cs.HC ·2026-10-22 ·Aleix Sant, Jordi Luque, Fernando López, Pablo Gómez, Carlos Escolano
Abstract

Everyday household information is often exchanged in conversations but never written down, making details easy to forget or misremember. We present a voice-enabled smart-home assistant that acts as a household memory by recording conversations and retrieving information from stored and newly captured speech. The assistant runs on an Android tablet with a modular architecture in which speech-recognition, diarisation, embedding and language models can be configured or replaced through an orchestration layer. On-device wake-word detection and VAD-based query capture support hands-free interaction. Recorded audio is processed in configurable chunks, transcribed and diarised, converted into turn-aware passages with conversation-level timestamps, and indexed in a vector store. For memory queries, a retrieval-augmented generation pipeline selects key passages and uses them to generate an answer. The interface supports recording, conversation management and inspection of timestamped sources.

0.22

Adaptive Uncertainty Visualization With jk-Plots

cs.HC ·2026-10-22 ·Nikolaus Piccolotto, Daniel Pahr, Laura Lotteraner, Fatih Öztank, Markus Bögl
Abstract · 5 pages

An important task of uncertainty visualizations is to communicate how the number of available observations influences the precision of estimates, such as a Normal distribution's mean. In this paper, we propose an adaptive approach to visualizing univariate data that produces salient representations while highlighting the uncertainty introduced by the sample size. We introduce jk-plots, inspired by educational material of Bayesian statistics, that combine visualizations of plausible reference models (emulating popular uncertainty visualization idioms) and the observations. We describe a visualization pipeline to create jk-plots, which we implemented in an R package. We demonstrate jk-plots using synthetic datasets of varying size and discuss possible usage scenarios in domain expert interviews.

0.22

Atlas: Making Sense of Long Horizon Agent Trajectories

cs.HC ·2026-10-22 ·Sangho Suh, Joseph Chee Chang, Pao Siangliulue, Amy X. Zhang, Matt Latzke +3 more
Abstract

AI agents can now search literature, generate hypotheses, analyze datasets, and run experiments at a scale difficult for researchers to match. Yet scaling scientific execution does not automatically scale researchers' ability to make sense of what agents produce. When researchers return to a long-horizon run, they must evaluate work they did not monitor: what the agent produced, which decisions shaped those outputs, and whether those decisions served the research goal. We frame this post-run sensemaking as a representation problem across three linked levels: goal, decision, and artifact. From this framework we developed Atlas, an agentic workspace with a trajectory tree, which renders the run as a branching structure showing what the agent chose and set aside, and artifact projections over six representations. In a within-subjects study (N=14), participants reported higher decision transparency, artifact sensemaking, and goal connection with Atlas than with a standard baseline, together with substantially lower subjective workload.

0.21

CARL: Chat Assisted Reinforcement Learning

cs.HC ·2026-10-22 ·Sanjiv Das, Harshad Khadilkar, Sukrit Mittal, Daniel Ostrov, Jacob Sisk
Abstract · Revision fixed citation format

Failure of applied Reinforcement Learning (RL) to solve a given problem predominantly stems from unaddressed specification gaps, where the intent of human experts in the problem's domain is inaccurately or incompletely translated by AI into mathematical objectives, constraints, and reward functions. Failure can also stem from unaddressed performance gaps, where human domain experts are unable to determine the validity of the RL based solution or where parameters (i.e., hyperparameters or weights on the reward terms comprising the reward) are poorly chosen. Chat Assisted Reinforcement Learning (CARL) bridges these gaps through a human-AI co-design interface architecture that enforces interactive specification prior to code generation and then enables subsequent verification and parameter alterations via iteration. More specifically, CARL leads domain experts through a structured ten-step workflow for each iteration using step-specific skill files, domain-aware suggestions via an assistant persona (called Trillian), and compact decision summaries that maintain context across iterations without context overload. Once the domain expert confirms the Markov decision process (MDP) specified by this procedure, CARL's autonomous tool-use loop generates, tests, executes, and debugs an end-to-end RL training pipeline, followed by an interactive post-training diagnosis and revisit step that allows for verification and parameter alteration before running a new iteration if desired.

0.17

Beyond Better Ideas: How Human-AI Collaboration Shapes Scientific Ideas and Human Perceptions

cs.HC ·2026-10-22 ·Qian Chen, Saatvik Pradhan, Janice Ahn, Berk Atil, Angkai Li +5 more
Abstract · 23 pages, 10 figures, 9 tables

As AI increasingly becomes a collaborator in scientific research, its impact may extend beyond improving ideas to reshaping how they are evaluated, attributed, and perceived by humans. Yet existing research has largely examined these dimensions in isolation. We present a unified empirical study of Human-AI scientific ideation, investigating two complementary perspectives: i) how AI collaboration shapes scientific ideas and their evaluation, and ii) how firsthand collaboration reshapes human perceptions of AI. Through a shared experimental setting involving Human-AI and Human-Human research teams, human and LLM evaluators, collaboration-provenance analysis, and participant surveys, we uncover several counterintuitive findings. First, AI collaboration shows no significant advantage in human-evaluated idea quality, while iterative revision consistently improves quality. Second, AI judges can approximate human scores yet disagree on idea rankings; notably, an AI judge favors AI-assisted proposals without a corresponding advantage in human evaluations. Third, collaboration provenance remains only partially detectable, revealing substantial overlap between Human-AI and Human-Human scientific outputs. Finally, firsthand AI collaboration is associated with greater perceived usefulness but smaller efficiency gains than expected, suggesting a nuanced recalibration of human expectations. Overall, our findings challenge the notion that AI's value in scientific ideation lies primarily in producing better ideas, revealing instead its broader influence on scientific creation, evaluation, attribution, and human perceptions.

0.10

AI Risk Ontologies for the EU AI Act Compliance: AIRO, VAIR, and their Integration with DPV

cs.CY cs.AI ·2026-10-22 ·Delaram Golpayegani, Harshvardhan J. Pandit, Declan O'Sullivan, Dave Lewis
Abstract

As the use of AI, and consequently its harms, becomes more prevalent, the global discourse around AI has become increasingly risk-aware. With new risks emerging more frequently than ever, a key challenge is how to represent, document, and catalogue AI risks in an effective, interoperable, and queryable manner. Additionally, the European Union (EU) AI Act's requirements for "high-risk" AI systems to manage and document the risks creates the need for open and interoperable tools, methods, and standards that facilitate AI risk management in an ever-changing environment. In this paper, we present ontologies and taxonomies developed to assist with risk management and documentation, in line with the requirements of the EU the AI Act. We developed the AI Risk Ontology (AIRO) to specify AI use cases and their associated risks, complemented by the Vocabulary of AI Risks (VAIR), which populates AIRO with taxonomies of concepts. Key international standards on AI and risk, namely ISO/IEC 23894 on AI risk management and ISO/IEC 22989 on AI terminology informed the development of the ontologies. To promote reuse and enable multi-regulatory compliance, AIRO and VAIR have been integrated in the Data Privacy Vocabulary (DPV) through a consensus-building process, expanding DPV's scope from data to AI.

0.09

Major Revision, Minor Accountability: LLMs as Reviewers of Published Articles in Top-Tier Journals

cs.DL cs.AI ·2026-10-22 ·Bastien Le Guellec, Su Hwan Kim, Raphael Bentegeac, Aghiles Hamroun
Abstract

With the growth of scientific output, the peer review system is approaching a breaking point: reviewers are increasingly hard to secure, and turnaround times lengthen accordingly. LLM-assisted or LLM-generated peer review has been proposed as a remedy, but how such models would treat manuscripts already published in top-tier journals, the closest thing the current system has to a gold standard of scientific validity, is not known. Here we assembled a corpus of 200 articles, each released under a CC BY 4.0 license and published in 2026, from 20 general-science, life-science, and clinical-medicine journals including Nature, Science, and Cell, and asked four current LLMs (two proprietary, two open-weight) to produce independent peer-review reports of each one under a blinded protocol. Their decisions were overwhelmingly harsh and starkly inconsistent with each other: 605 of 800 reviews (75.6%) recommended major revision or rejection, but the rate at which any single model did so ranged from 17.5% to 100% (p < 0.001), disagreement on manuscripts that had, by construction, already survived real editorial review at the journals in question. When the same manuscripts were doctored with one deliberately planted, field-specific methodological or statistical flaw, detection rates varied from 17.0% to 60.5% across models (p < 0.001). Recommending revision for papers that had already passed editorial review (low specificity) while missing flaws deliberately placed in front of them (low sensitivity), these models produce output that should be read for what it is: an uneven, framing- and model-dependent signal.

0.07