CS Paper Radar

Variance Reduction for Decentralized Stochastic Optimization with Multiple Local Updates

Jie Liu, Nicola Bastianello, Wenying Xu, Daniel W. C. Ho, Karl H. Johansson

cs.LG math.OC ·submitted 2026-10-30 15:54 UTC · v2

Abstract

Multiple local updates are widely used in decentralized optimization to enhance communication efficiency, often in conjunction with stochastic gradients when exact gradient computation is costly. However, such algorithms generally converge only to a neighborhood of the optimum, whose size depends on the stepsize and the variance of gradient noise. Variance reduction has been incorporated into ADMM-based methods to achieve exact convergence in expectation, but their stepsizes typically scale inversely with both the square root of the network size and the cube of the number of local updates, leading to slow convergence as the network size or the number of local updates increases. In this paper, we propose a novel primal-dual-based decentralized optimization algorithm that integrates variance reduction with multiple local updates, with a stepsize independent of the network size and inversely proportional to the number of local updates (rather than its cube), thereby enabling significantly larger stepsizes and faster convergence. Moreover, to the best of our knowledge, we establish the first linear convergence guarantee in expectation for decentralized optimization with multiple local updates under stochastic gradients in the strongly convex setting. Beyond the theoretical guarantee, the proposed method exchanges only one variable per round, halving the communication cost compared with gradient-tracking-based methods. Numerical experiments further demonstrate its advantages in convergence speed and accuracy.

Closest papers

Nearest neighbours by meaning across the whole index.

Accelerated Push-Pull with Local Updates for Decentralized Stochastic Optimization

eess.SY math.OC ·2026-10-26 ·Sheng Dong, Weiguo Xia, Huijie Tao
theory similarity 0.71
Abstract · 38 pages, 4 figures

We consider decentralized stochastic nonconvex optimization over directed networks. Push--Pull provides an efficient decentralized optimization framework for such networks, but in the stochastic setting, gradient variance can slow its convergence and complicate the analysis of linear speedup. Motivated by this observation, we incorporate recursive momentum into stochastic Push--Pull for variance reduction, yielding the accelerated stochastic Push--Pull (ASPP) algorithm. Under mean-squared smoothness, we prove that ASPP improves the convergence rate of stochastic Push--Pull and achieves asymptotic linear speedup with respect to the number of agents. To further reduce communication, we develop LASPP, a local-update variant of ASPP that allows each agent to perform multiple local updates between communication rounds. We show that LASPP achieves an additional asymptotic linear speedup with respect to the number of local steps. With a suitable number of local steps, LASPP reduces communication complexity while maintaining the same stochastic gradient complexity as ASPP. Numerical experiments on real datasets demonstrate the effectiveness of the proposed algorithms.

0.25

Decentralized SGD under Heavy-Tailed Noise: Optimal Convergence Rates and the Role of Gradient Clipping

cs.LG math.OCcs.MA ·2026-10-07 ·Aleksandar Armacki, Haoyuan Cai, Ali H. Sayed
similarity 0.68
Abstract · 36 pages, 5 figures, 2 tables

Heavy-tailed noise has been widely observed in modern machine learning, motivating the use of methods like gradient clipping and normalization. While these methods are well understood in centralized settings, much less is known in decentralized ones, where applying a nonlinearity to local gradients affects both optimization and consensus. Recent works on decentralized non-convex optimization have studied both clipping and normalization under heavy-tailed noise, with clipping yielding suboptimal rates and normalization needing local momentum or mini-batches to converge. This raises the question: can a baseline decentralized method using a nonlinearity achieve optimal convergence rates under heavy-tailed noise? We answer affirmatively with clipped decentralized SGD ($\mathtt{DSGD}$). For smooth non-convex costs under bounded $p$-th moment noise, $p \in (1,2]$, we show that clipped $\mathtt{DSGD}$ achieves order-optimal rates both with high probability and in expectation. Moreover, we establish a linear speed-up in the number of agents, which, to our knowledge, has not been shown for decentralized methods with clipping. The key technical ingredient is a sharp analysis of the consensus gap that exploits the structure of clipping, relegating network effects to higher-order terms. Our results highlight an important distinction between clipping and normalization in decentralized settings: while normalized $\mathtt{DSGD}$ can fail to converge, clipping retains magnitude information, enabling $\mathtt{DSGD}$ to be convergent and order-optimal. Numerical experiments validate our theory.

0.42

Distributed Zeroth-Order Composite Finite-Sum Optimization via Incremental Variance Reduction

eess.SY math.OC ·2026-10-30 ·Silan Zhang, Yujie Tang
similarity 0.66
Abstract

We study decentralized composite finite-sum optimization over a fixed connected undirected network, where agents access local component functions through function evaluations and share a possibly nonsmooth convex regularizer. In this setting, standard two-point gradient estimators can retain nonvanishing variance at a solution, complicating their integration with gradient tracking. We propose Distributed Zeroth-Order Incremental Variance Reduction (Distributed ZIVR), which combines incrementally updated component-gradient estimates with a proximal gradient-tracking scheme. The method admits an implementation using a constant number of function evaluations per agent at every iteration, without intermittent full-component or full-coordinate refreshes. For convex smooth component functions and strongly convex local averages, we establish linear convergence up to a controllable finite-difference error. With the finite-difference radius chosen for the target accuracy, Chebyshev-accelerated mixing yields a per-agent oracle complexity of $\widetilde{O}(d(m+κ)\log(1/ε))$ for expected squared solution error $ε$, where $d$ is the dimension, $m$ is the number of components per agent, and $κ$ is the condition number. This matches the centralized ZIVR oracle bound, while the communication bound has only an inverse square-root dependence on the network spectral gap. Experiments on logistic and robust regression demonstrate rapid reduction of solution and consensus errors.

0.13

Revisiting Distributed Sign-Based Variance Reduction

cs.LG math.OCstat.ML ·2026-09-16 ·Wei Jiang, Zechao Li, Lijun Zhang
similarity 0.66
Abstract

Sign-based methods reduce communication costs in distributed environments, but aggregating local signs can introduce bias when data are heterogeneous. As a result, existing sign-based variance reduction methods fail to obtain the optimal convergence rates. In this paper, we solve this problem and obtain optimal rates for both nonconvex stochastic and finite-sum optimization. We first give a counterexample showing that majority voting can fail to approach stationary points even with exact local gradients. Motivated by this limitation, we propose tracking the global gradient at the server through unbiased compression of recursive gradient increments. As a result, we can obtain the convergence rates of $O(\sqrt{d/K}+\sqrt d (a/(nK))^{1/3})$ for the $\ell_1$-norm and $O(\sqrt{a/K}+\sqrt a/(nK)^{1/3})$ for the $\ell_2$-norm. Here, $K$ is the iteration number, $n$ is the number of workers, $d$ is the dimension, and $a=1+ω$, with $ω$ denoting the compressor's relative variance. For finite-sum problems with $M$ components, we combine periodic exact gradient refreshes with compressed component-gradient differences. The resulting total sample complexities are $O(M+d\sqrt{aM}ε^{-2})$ and $O(M+a\sqrt M\ epsilon^{-2})$ for $\ell_1$ and $\ell_2$ gradient norms at most $ε$, matching the corresponding bounds in centralized settings.

0.55

KKT-Perturbation-Aware Communication for Fully Distributed Online Convex Optimization with Nonseparable Coupling

cs.LG ·2026-11-09 ·Zhaoye Pan, Tao Li, Yunjie Guo, Guanpu Chen
similarity 0.64
Abstract

Nonseparable costs and coupled constraints in distributed online convex optimization often arise when system performance and operational constraints depend on the joint configuration of all agents and cannot be decomposed into individual contributions. Such nonseparability leads to fully distributed decision-making with challenging partial local information, where each agent evaluates its local primal-dual update using estimated global information via a local belief of the joint action and the shared Lagrange multiplier to be exchanged with neighbors. While the belief consensus established by prior works ensures no-regret learning, exchanging beliefs comprising full-dimensional primal and dual states at every round imposes a communication overhead that scales with both the joint decision dimension and the network size. Existing event-triggered schemes typically activate communication according to state variation or consensus error. Under nonseparable coupling, however, errors of equal magnitude can produce markedly different perturbations after passing through the KKT operator. We therefore propose a randomized KKT-perturbation-aware communication mechanism, where the triggering of each agent's primal and dual communication channels depends on their impact on the KKT operator. The proposed mechanism transfers the adaptive-sampling principle of coordinate descent from allocating computation across decision blocks to allocating communication across information channels: channels with larger estimated perturbations receive greater priority for refresh. Under standard regularity assumptions, the resulting algorithm achieves $O({T}^{2/3})$ expected static regret and $O({T}^{2/3})$ expected cumulative constraint violation. Numerical experiments demonstrate improved learning and constraint-enforcement performance per transmitted bit over belief consensus baselines.

0.19

Group Distributionally Robust Optimization over Decentralized Networks

cs.LG math.OC ·2026-10-21 ·Haomin Bai, Junyan Sun, Quan Cheng, Yuanyu Wan, Lijun Zhang
similarity 0.62
Abstract

We study group distributionally robust optimization (GDRO) over decentralized networks, where workers collaborate to learn robust models by minimizing the maximum local risk. Existing methods optimize only dual-regularized surrogates and establish guarantees for a globally averaged model that is not available to workers. In this paper, we directly optimize the original GDRO objective and establish guarantees for local models in semi-decentralized (SemiDec) and fully decentralized (Dec) settings. A key difficulty is that applying the primal-dual decomposition from centralized GDRO yields worker-dependent losses whose evaluation requires unavailable cross-worker information. To address this challenge, we develop a novel error decomposition that recasts decentralized GDRO as (i) a decentralized online convex optimization (D-OCO) problem and a prediction-with-expert-advice problem in SemiDec, and as (ii) two coupled D-OCO problems in Dec, with discrepancy terms capturing inconsistencies among local decisions. By isolating consensus errors in analysis, we develop modular frameworks that support different consensus routines. We then establish the \emph{first} per-worker convergence guarantees for optimizing the original GDRO objective over decentralized networks. Instantiations with accelerated gossip and Choco-gossip achieve an $O(\sqrt{(\log m)/T})$ convergence rate over $T$ update rounds, while experiments demonstrate improved worst-worker performance over baselines.

0.36

Budget-Constrained Multi-Consensus Decentralized Gradient Descent

eess.SY math.OC ·2026-10-06 ·Shuyi Ren, Nicol`o Michelusi, Erik G. Larsson
hot direction similarity 0.62
Abstract

We investigate decentralized gradient descent (DGD) with emphasis on efficient communication and computation resource utilization under budget constraints. As a first step toward the broader communication-computation allocation problem, we consider and analyze a \textit{multi-consensus decentralized gradient descent} (mcDGD) scheme, where the number of consensus rounds and the stepsize are allowed to vary across iterations. Building on a unified analytical framework for DGD, we derive finite-time convergence bounds that explicitly characterize the interaction between consensus quality and optimization dynamics. Our analysis requires only convexity of the local objective functions while assuming smoothness and strong convexity of the global objective. The resulting bounds enable a principled consensus-allocation strategy under resource constraints, for which we show that equal allocation of consensus rounds across iterations is optimal under our stepsize rule, up to integer rounding. Numerical experiments corroborate the theoretical findings and demonstrate favorable communication-computation tradeoffs compared with existing multi-consensus decentralized optimization baselines.

0.30

DisMUSE: An Efficient Distributed Singular-Spectrum-Based Optimization Approach with Theoretical Convergence Guarantee

cs.LG math.OC ·2026-11-06 ·Zhen Qin, Zhiyao Zhang, Yingbin Liang, Jia Liu
similarity 0.62
Abstract

In recent years, singular-spectrum-based optimizers have attracted increasing attention due to their ability to preserve the matrix structure of neural network gradients, thereby enabling geometry-aware optimization. However, most singular-spectrum-based (SSB) approaches remain centralized, limiting their use in distributed learning infrastructures that exploit massive parallelism. Distributed SSB optimization is still at an early stage, with open questions about convergence guarantees and communication efficiency. To fill this gap, we propose the Distributed Momentum-based Spectral Estimator (DisMUSE), an efficient distributed SSB optimization framework for both data-center and federated settings. Using Muon-style exact orthogonalized momentum as the spectral update, DisMUSE integrates a recursive variance-reduced momentum estimator and achieves an $O((KT)^{-1/3})$ convergence rate, where $K$ and $T$ denote the number of nodes and local iterations per node, respectively. This rate matches the best-known rates for variance-reduced nonconvex distributed optimization. Building on this variance-reduced framework, we study two extensions. First, our DisMUSE method with finite Newton-Schulz (NS) iterations preserves the same convergence order under a uniform normalized support-spectrum condition, up to a multiplicative factor. Second, we develop a communication-efficient variant based on low-rank (LR) momentum transmission and quantify the error it introduces to the stationarity bound, characterizing the trade-off between accuracy and communication cost.

0.16