State-based Episodic Memory for Multi-Agent Reinforcement Learning

TL;DR

This paper introduces state-based episodic memory (SEM) to enhance sample efficiency in multi-agent reinforcement learning by integrating episodic memory into the training process, reducing computational costs and improving learning performance.

Contribution

The paper is the first to incorporate episodic memory into multi-agent reinforcement learning, demonstrating theoretical advantages and empirical improvements over existing methods.

Findings

SEM improves sample efficiency in MARL on SMAC.

SEM reduces storage and time costs compared to SAEM.

Theoretical analysis shows SEM has lower complexity than SAEM.

Abstract

Multi-agent reinforcement learning (MARL) algorithms have made promising progress in recent years by leveraging the centralized training and decentralized execution (CTDE) paradigm. However, existing MARL algorithms still suffer from the sample inefficiency problem. In this paper, we propose a simple yet effective approach, called state-based episodic memory (SEM), to improve sample efficiency in MARL. SEM adopts episodic memory (EM) to supervise the centralized training procedure of CTDE in MARL. To the best of our knowledge, SEM is the first work to introduce EM into MARL. We can theoretically prove that, when using for MARL, SEM has lower space complexity and time complexity than state and action based EM (SAEM), which is originally proposed for single-agent reinforcement learning. Experimental results on StarCraft multi-agent challenge (SMAC) show that introducing episodic memory into MARL can improve sample efficiency and SEM can reduce storage cost and time cost compared with SAEM.