English

Cooperative Multi-Agent Reinforcement Learning: Asynchronous Communication and Linear Function Approximation

Machine Learning 2023-06-27 v3 Artificial Intelligence Multiagent Systems

Abstract

We study multi-agent reinforcement learning in the setting of episodic Markov decision processes, where multiple agents cooperate via communication through a central server. We propose a provably efficient algorithm based on value iteration that enable asynchronous communication while ensuring the advantage of cooperation with low communication overhead. With linear function approximation, we prove that our algorithm enjoys an O~(d3/2H2K)\tilde{\mathcal{O}}(d^{3/2}H^2\sqrt{K}) regret with O~(dHM2)\tilde{\mathcal{O}}(dHM^2) communication complexity, where dd is the feature dimension, HH is the horizon length, MM is the total number of agents, and KK is the total number of episodes. We also provide a lower bound showing that a minimal Ω(dM)\Omega(dM) communication complexity is required to improve the performance through collaboration.

Keywords

Cite

@article{arxiv.2305.06446,
  title  = {Cooperative Multi-Agent Reinforcement Learning: Asynchronous Communication and Linear Function Approximation},
  author = {Yifei Min and Jiafan He and Tianhao Wang and Quanquan Gu},
  journal= {arXiv preprint arXiv:2305.06446},
  year   = {2023}
}

Comments

Published at the 40th International Conference on Machine Learning ( ICML 2023 )