Provably Efficient Offline Multi-agent Reinforcement Learning via Strategy-wise Bonus
share
This paper considers offline multi-agent reinforcement learning. We propose the strategy-wise concentration principle which directly builds a confidence interval for the joint strategy, in contrast to the point-wise concentration principle that builds a confidence interval for each point in the joint action space. For two-player zero-sum Markov games, by exploiting the convexity of the strategy-wise bonus, we propose a computationally efficient algorithm whose sample complexity enjoys a better dependency on the number of actions than the prior methods based on the point-wise bonus. Furthermore, for offline multi-agent general-sum Markov games, based on the strategy-wise bonus and a novel surrogate function, we give the first algorithm whose sample complexity only scales ∑_i=1^mA_i where A_i is the action size of the i-th player and m is the number of players. In sharp contrast, the sample complexity of methods based on the point-wise bonus would scale with the size of the joint action space Π_i=1^m A_i due to the curse of multiagents. Lastly, all of our algorithms can naturally take a pre-specified strategy class Π as input and output a strategy that is close to the best strategy in Π. In this setting, the sample complexity only scales with log |Π| instead of ∑_i=1^mA_i.
READ FULL TEXT