On the Optimality of Batch Policy Optimization Algorithms

Abstract

Batch policy optimization considers leveraging existing data for policy construction before interacting with an environment. Although interest in this problem has grown significantly in recent years, its theoretical foundations remain under-developed. To advance the understanding of this problem, we provide three results that characterize the limits and possibilities of batch policy optimization in the finite-armed stochastic bandit setting. First, we introduce a class of confidence-adjusted index algorithms that unifies optimistic and pessimistic principles in a common framework, which enables a general analysis. For this family, we show that any confidence-adjusted index algorithm is minimax optimal, whether it be optimistic, pessimistic or neutral. Our analysis reveals that instance-dependent optimality, commonly used to establish optimality of on-line stochastic bandit algorithms, cannot be achieved by any algorithm in the batch setting. In particular, for any algorithm that performs optimally in some environment, there exists another environment where the same algorithm suffers arbitrarily larger regret. Therefore, to establish a framework for distinguishing algorithms, we introduce a new weighted-minimax criterion that considers the inherent difficulty of optimal value prediction. We demonstrate how this criterion can be used to justify commonly used pessimistic principles for batch policy optimization.

arXiv comment29 pages, 8 figures

References in corpus (10)

• Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems
• Way Off-Policy Batch Deep Reinforcement Learning of Implicit Human Preferences in Dialog
• Learning from Logged Implicit Exploration Data
• Keep Doing What Worked: Behavioral Modelling Priors for Offline Reinforcement Learning
• Provably Good Batch Reinforcement Learning Without Great Exploration
• The Importance of Pessimism in Fixed-Dataset Policy Optimization
• CoinDICE: Off-Policy Confidence Interval Estimation
• Near-Optimal Offline Reinforcement Learning via Double Variance Reduction
• Distributional Robustness and Regularization in Reinforcement Learning
• Empirical Likelihood for Contextual Bandits

Cited by in corpus (2)

• Towards Theoretical Understandings of Robust Markov Decision Processes: Sample Complexity and Asymptotics
• Offline Neural Contextual Bandits: Pessimism, Optimization and Generalization