QPLEX: Duplex Dueling Multi-Agent Q-Learning
This addresses scalability and stability issues in multi-agent systems for complex domains like gaming, though it is an incremental improvement over existing CTDE methods.
The paper tackles the challenge of maintaining Individual-Global-Max consistency in multi-agent reinforcement learning by introducing QPLEX, a duplex dueling network architecture that factors the joint value function, resulting in significant outperformance over state-of-the-art baselines on StarCraft II tasks in both online and offline settings.
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.