Soft-Robust Algorithms for Batch Reinforcement Learning

Abstract

In reinforcement learning, robust policies for high-stakes decision-making problems with limited data are usually computed by optimizing the percentile criterion, which minimizes the probability of a catastrophic failure. Unfortunately, such policies are typically overly conservative as the percentile criterion is non-convex, difficult to optimize, and ignores the mean performance. To overcome these shortcomings, we study the soft-robust criterion, which uses risk measures to balance the mean and percentile criterion better. In this paper, we establish the soft-robust criterion's fundamental properties, show that it is NP-hard to optimize, and propose and analyze two algorithms to approximately optimize it. Our theoretical analyses and empirical evaluations demonstrate that our algorithms compute much less conservative solutions than the existing approximate methods for optimizing the percentile-criterion.

References in corpus (7)

• Continuous control with deep reinforcement learning
• Deep Q-learning from Demonstrations
• Algorithms for CVaR Optimization in MDPs
• Robust Reinforcement Learning for Continuous Control with Model Misspecification
• Policy Gradient for Coherent Risk Measures
• Lightning Does Not Strike Twice: Robust MDPs with Coupled Uncertainty
• Soft-Robust Actor-Critic Policy-Gradient