Towards General Function Approximation in Zero-Sum Markov Games
arXiv:2107.14702
Abstract
This paper considers two-player zero-sum finite-horizon Markov games with simultaneous moves. The study focuses on the challenging settings where the value function or the model is parameterized by general function classes. Provably efficient algorithms for both decoupled and {coordinated} settings are developed. In the {decoupled} setting where the agent controls a single player and plays against an arbitrary opponent, we propose a new model-free algorithm. The sample complexity is governed by the Minimax Eluder dimension -- a new dimension of the function class in Markov games. As a special case, this method improves the state-of-the-art algorithm by a factor in the regret when the reward function and transition kernel are parameterized with -dimensional linear features. In the {coordinated} setting where both players are controlled by the agent, we propose a model-based algorithm and a model-free algorithm. In the model-based algorithm, we prove that sample complexity can be bounded by a generalization of Witness rank to Markov games. The model-free algorithm enjoys a -regret upper bound where is the number of episodes.
References in corpus (9)
- Safe, Multi-Agent, Reinforcement Learning for Autonomous Driving
- Contextual Decision Processes with Low Bellman Rank are PAC-Learnable
- Contextual Bandit Learning with Predictable Rewards
- Model-based Reinforcement Learning and the Eluder Dimension
- Reinforcement Learning with General Value Function Approximation: Provably Efficient Approach via Bounded Eluder Dimension
- Feature-Based Q-Learning for Two-Player Stochastic Games
- On Function Approximation in Reinforcement Learning: Optimism in the Face of Large State Spaces
- Near-Optimal Reinforcement Learning with Self-Play
- The Power of Exploiter: Provable Multi-Agent RL in Large State Spaces