Finite Sample Analysis of Minimax Offline Reinforcement Learning: Completeness, Fast Rates and First-Order Efficiency

Abstract

We offer a theoretical characterization of off-policy evaluation (OPE) in reinforcement learning using function approximation for marginal importance weights and $q$-functions when these are estimated using recent minimax methods. Under various combinations of realizability and completeness assumptions, we show that the minimax approach enables us to achieve a fast rate of convergence for weights and quality functions, characterized by the critical inequality \citep{bartlett2005}. Based on this result, we analyze convergence rates for OPE. In particular, we introduce novel alternative completeness conditions under which OPE is feasible and we present the first finite-sample result with first-order efficiency in non-tabular environments, i.e., having the minimal coefficient in the leading term.

Under Review

References in corpus (13)

• AlgaeDICE: Policy Gradient from Arbitrary Experience
• Information-Theoretic Considerations in Batch Reinforcement Learning
• GenDICE: Generalized Offline Estimation of Stationary Values
• What are the Statistical Limits of Offline RL with Linear Function Approximation?
• A unifying approach for doubly-robust ${\ell }_1$ regularized estimation of causal contrasts
• Near-Optimal Provable Uniform Convergence in Offline Policy Evaluation for Reinforcement Learning
• Minimax Estimation of Conditional Moment Models
• Off-Policy Evaluation via the Regularized Lagrangian
• More Efficient Off-Policy Evaluation through Regularized Targeted Learning
• Minimax Value Interval for Off-Policy Evaluation and Policy Optimization
• Instrumental Variable Regression via Kernel Maximum Moment Loss
• Sparse Feature Selection Makes Batch Reinforcement Learning More Sample Efficient
• A Variant of the Wang-Foster-Kakade Lower Bound for the Discounted Setting

Cited by in corpus (5)

• Mitigating Covariate Shift in Imitation Learning via Offline Data Without Great Coverage
• Pessimistic Model-based Offline Reinforcement Learning under Partial Coverage
• Infinite-Horizon Offline Reinforcement Learning with Linear Function Approximation: Curse of Dimensionality and Algorithm
• Offline Reinforcement Learning: Fundamental Barriers for Value Function Approximation
• Pessimistic Model Selection for Offline Deep Reinforcement Learning