Value Function Approximation in Zero-Sum Markov Games

Abstract

This paper investigates value function approximation in the context of zero-sum Markov games, which can be viewed as a generalization of the Markov decision process (MDP) framework to the two-agent case. We generalize error bounds from MDPs to Markov games and describe generalizations of reinforcement learning algorithms to Markov games. We present a generalization of the optimal stopping problem to a two-player simultaneous move Markov game. For this special problem, we provide stronger bounds and can guarantee convergence for LSTD and temporal difference learning with linear value function approximation. We demonstrate the viability of value function approximation for Markov games by using the Least squares policy iteration (LSPI) algorithm to learn good policies for a soccer domain and a flow control problem.

Appears in Proceedings of the Eighteenth Conference on Uncertainty in Artificial Intelligence (UAI2002)

References in corpus (1)

• Policy Iteration for Factored MDPs

Cited by in corpus (8)

• Multi-agent Reinforcement Learning in Sequential Social Dilemmas
• A Theoretical Analysis of Deep Q-Learning
• Actor-Critic Provably Finds Nash Equilibria of Linear-Quadratic Mean-Field Games
• From Poincaré Recurrence to Convergence in Imperfect Information Games: Finding Equilibrium via Regularization
• Learning Nash Equilibrium for General-Sum Markov Games from Batch Data
• Combining Pessimism with Optimism for Robust and Efficient Model-Based Deep Reinforcement Learning
• On Reinforcement Learning for Turn-based Zero-sum Markov Games
• Learning Nash Equilibria in Zero-Sum Stochastic Games via Entropy-Regularized Policy Approximation