Preference-Based Multi-Agent Reinforcement Learning: Data Coverage and Algorithmic Techniques

TL;DR

This paper studies Preference-Based Multi-Agent Reinforcement Learning, establishing theoretical complexity bounds, and introduces novel algorithms to improve reward learning and training stability in preference-only offline datasets.

Contribution

It provides the first theoretical analysis of PbMARL's complexity and proposes two new algorithmic techniques for better reward distribution and training stability.

Findings

Single-policy coverage is insufficient for Nash equilibrium identification.

Unilateral dataset coverage is crucial for effective PbMARL.

Proposed algorithms improve reward learning and training stability.

Abstract

We initiate the study of Preference-Based Multi-Agent Reinforcement Learning (PbMARL), exploring both theoretical foundations and empirical validations. We define the task as identifying the Nash equilibrium from a preference-only offline dataset in general-sum games, a problem marked by the challenge of sparse feedback signals. Our theory establishes the upper complexity bounds for Nash Equilibrium in effective PbMARL, demonstrating that single-policy coverage is inadequate and highlighting the importance of unilateral dataset coverage. These theoretical insights are verified through comprehensive experiments. To enhance the practical performance, we further introduce two algorithmic techniques. (1) We propose a Mean Squared Error (MSE) regularization along the time axis to achieve a more uniform reward distribution and improve reward learning outcomes. (2) We propose an additional penalty based on the distribution of the dataset to incorporate pessimism, improving stability and effectiveness during training. Our findings underscore the multifaceted approach required for PbMARL, paving the way for effective preference-based multi-agent systems.