Model-based Multi-agent Policy Optimization with Adaptive Opponent-wise Rollouts
This work addresses sample efficiency for multi-agent reinforcement learning practitioners, but it is incremental as it builds on existing model-based methods with specific theoretical and empirical improvements.
This paper tackles the sample complexity problem in multi-agent reinforcement learning by proposing a decentralized model-based method called AORPO, which reduces the return discrepancy upper bound and achieves improved sample efficiency with comparable asymptotic performance in competitive and cooperative tasks.
This paper investigates the model-based methods in multi-agent reinforcement learning (MARL). We specify the dynamics sample complexity and the opponent sample complexity in MARL, and conduct a theoretic analysis of return discrepancy upper bound. To reduce the upper bound with the intention of low sample complexity during the whole learning process, we propose a novel decentralized model-based MARL method, named Adaptive Opponent-wise Rollout Policy Optimization (AORPO). In AORPO, each agent builds its multi-agent environment model, consisting of a dynamics model and multiple opponent models, and trains its policy with the adaptive opponent-wise rollout. We further prove the theoretic convergence of AORPO under reasonable assumptions. Empirical experiments on competitive and cooperative tasks demonstrate that AORPO can achieve improved sample efficiency with comparable asymptotic performance over the compared MARL methods.