English

Multi-Agent Reinforcement Learning via Double Averaging Primal-Dual Optimization

Machine Learning 2019-01-10 v4 Optimization and Control Machine Learning

Abstract

Despite the success of single-agent reinforcement learning, multi-agent reinforcement learning (MARL) remains challenging due to complex interactions between agents. Motivated by decentralized applications such as sensor networks, swarm robotics, and power grids, we study policy evaluation in MARL, where agents with jointly observed state-action pairs and private local rewards collaborate to learn the value of a given policy. In this paper, we propose a double averaging scheme, where each agent iteratively performs averaging over both space and time to incorporate neighboring gradient information and local reward information, respectively. We prove that the proposed algorithm converges to the optimal solution at a global geometric rate. In particular, such an algorithm is built upon a primal-dual reformulation of the mean squared projected Bellman error minimization problem, which gives rise to a decentralized convex-concave saddle-point problem. To the best of our knowledge, the proposed double averaging primal-dual optimization algorithm is the first to achieve fast finite-time convergence on decentralized convex-concave saddle-point problems.

Keywords

Cite

@article{arxiv.1806.00877,
  title  = {Multi-Agent Reinforcement Learning via Double Averaging Primal-Dual Optimization},
  author = {Hoi-To Wai and Zhuoran Yang and Zhaoran Wang and Mingyi Hong},
  journal= {arXiv preprint arXiv:1806.00877},
  year   = {2019}
}

Comments

final version as appeared in NeurIPS 2018

R2 v1 2026-06-23T02:17:33.661Z