Randomized Exploration in Cooperative Multi-Agent Reinforcement Learning

TL;DR

This paper introduces a new framework for randomized exploration in cooperative multi-agent reinforcement learning, providing the first theoretical guarantees and demonstrating improved empirical performance across diverse environments.

Contribution

It proposes a unified algorithm framework with two Thompson Sampling-based algorithms, CoopTS-PHE and CoopTS-LMC, achieving provable regret bounds in cooperative MARL.

Findings

Achieves a regret bound of (d^{3/2}H^2 ext{ extasciitilde} \sqrt{MK}) in linear transition MDPs.

Demonstrates superior performance on multiple RL benchmarks, including a deep exploration problem, video game, and energy system.

Establishes a connection between the framework and federated learning applications.

Abstract

We present the first study on provably efficient randomized exploration in cooperative multi-agent reinforcement learning (MARL). We propose a unified algorithm framework for randomized exploration in parallel Markov Decision Processes (MDPs), and two Thompson Sampling (TS)-type algorithms, CoopTS-PHE and CoopTS-LMC, incorporating the perturbed-history exploration (PHE) strategy and the Langevin Monte Carlo exploration (LMC) strategy, respectively, which are flexible in design and easy to implement in practice. For a special class of parallel MDPs where the transition is (approximately) linear, we theoretically prove that both CoopTS-PHE and CoopTS-LMC achieve a $\widetilde{\mathcal{O}}(d^{3/2}H^2\sqrt{MK})$ regret bound with communication complexity $\widetilde{\mathcal{O}}(dHM^2)$, where $d$ is the feature dimension, $H$ is the horizon length, $M$ is the number of agents, and $K$ is the number of episodes. This is the first theoretical result for randomized exploration in cooperative MARL. We evaluate our proposed method on multiple parallel RL environments, including a deep exploration problem (i.e., $N$-chain), a video game, and a real-world problem in energy systems. Our experimental results support that our framework can achieve better performance, even under conditions of misspecified transition models. Additionally, we establish a connection between our unified framework and the practical application of federated learning.