QPLEX: Duplex Dueling Multi-Agent Q-Learning

TL;DR

QPLEX introduces a duplex dueling network architecture for multi-agent Q-learning that enforces the IGM principle, leading to improved scalability, stability, and performance in complex multi-agent environments.

Contribution

It presents a novel duplex dueling network architecture that encodes the IGM principle into the neural network, enabling scalable and stable multi-agent reinforcement learning.

Findings

QPLEX outperforms state-of-the-art methods on StarCraft II tasks.

It achieves high sample efficiency and benefits from offline datasets.

Theoretical analysis confirms the complete IGM function class.

Abstract

We explore value-based multi-agent reinforcement learning (MARL) in the popular paradigm of centralized training with decentralized execution (CTDE). CTDE has an important concept, Individual-Global-Max (IGM) principle, which requires the consistency between joint and local action selections to support efficient local decision-making. However, in order to achieve scalability, existing MARL methods either limit representation expressiveness of their value function classes or relax the IGM consistency, which may suffer from instability risk or may not perform well in complex domains. This paper presents a novel MARL approach, called duPLEX dueling multi-agent Q-learning (QPLEX), which takes a duplex dueling network architecture to factorize the joint value function. This duplex dueling structure encodes the IGM principle into the neural network architecture and thus enables efficient value function learning. Theoretical analysis shows that QPLEX achieves a complete IGM function class. Empirical experiments on StarCraft II micromanagement tasks demonstrate that QPLEX significantly outperforms state-of-the-art baselines in both online and offline data collection settings, and also reveal that QPLEX achieves high sample efficiency and can benefit from offline datasets without additional online exploration.