The framework of centralized training with decentralized execution (CTDE) [8,28]isone ofthe popular frameworks for solving cooperative multi-agent tasks.

Despite the success of cooperative multi-agent reinforcement learning algorithms, most of them focus on a single team composition, which prevents them from being used in more realistic scenarios where dynamic team composition is possible. While some studies attempt to solve this problem via multi-task learning in a fixed set of team compositions, there is still a risk of overfitting to the training set, which may lead to catastrophic performance when facing dramatically varying team compositions during execution. To address this problem, we propose to use mutual information (MI) as an augmented reward to prevent individual policies from relying too much on team-related information and encourage agents to learn policies that are robust in different team compositions. Optimizing this MI-augmented objective in an off-policy manner can be intractable due to the existence of dynamic marginal distribution. To alleviate this problem, we first propose a multi-agent policy iteration algorithm with a fixed marginal distribution and prove its convergence and optimality. Then, we propose to employ the Blahut-Arimoto algorithm and an imaginary team composition distribution for optimization with approximate marginal distribution as the practical implementation. Empirically, our method demonstrates strong zero-shot generalization to dynamic team compositions in complex cooperative tasks.

The cooperative multi-agent reinforcement learning (MARL) problem has attracted the attention of many researchers for being a well-abstracted model for many real-world problems, such as traffic signal control (Wang et al., 2021), autonomous warehouse (Zhou et al., 2021), and even

Despite the success of cooperative multi-agent reinforcement learning algorithms, most of them focus on a single team composition, which prevents them from being used in more realistic scenarios where dynamic team composition is possible. While some studies attempt to solve this problem via multi-task learning in a fixed set of team compositions, there is still a risk of overfitting to the training set, which may lead to catastrophic performance when facing dramatically varying team compositions during execution. To address this problem, we propose to use mutual information (MI) as an augmented reward to prevent individual policies from relying too much on team-related information and encourage agents to learn policies that are robust in different team compositions. Optimizing this MI-augmented objective in an off-policy manner can be intractable due to the existence of dynamic marginal distribution. To alleviate this problem, we first propose a multi-agent policy iteration algorithm with a fixed marginal distribution and prove its convergence and optimality.