English

DASA: Delay-Adaptive Multi-Agent Stochastic Approximation

Artificial Intelligence 2024-08-05 v3 Robotics Systems and Control Systems and Control Optimization and Control Machine Learning

Abstract

We consider a setting in which NN agents aim to speedup a common Stochastic Approximation (SA) problem by acting in parallel and communicating with a central server. We assume that the up-link transmissions to the server are subject to asynchronous and potentially unbounded time-varying delays. To mitigate the effect of delays and stragglers while reaping the benefits of distributed computation, we propose \texttt{DASA}, a Delay-Adaptive algorithm for multi-agent Stochastic Approximation. We provide a finite-time analysis of \texttt{DASA} assuming that the agents' stochastic observation processes are independent Markov chains. Significantly advancing existing results, \texttt{DASA} is the first algorithm whose convergence rate depends only on the mixing time τmix\tau_{mix} and on the average delay τavg\tau_{avg} while jointly achieving an NN-fold convergence speedup under Markovian sampling. Our work is relevant for various SA applications, including multi-agent and distributed temporal difference (TD) learning, Q-learning and stochastic optimization with correlated data.

Keywords

Cite

@article{arxiv.2403.17247,
  title  = {DASA: Delay-Adaptive Multi-Agent Stochastic Approximation},
  author = {Nicolò Dal Fabbro and Arman Adibi and H. Vincent Poor and Sanjeev R. Kulkarni and Aritra Mitra and George J. Pappas},
  journal= {arXiv preprint arXiv:2403.17247},
  year   = {2024}
}
R2 v1 2026-06-28T15:33:28.188Z