Regret Bounds for Risk-Sensitive Reinforcement Learning

Osbert Bastani, Yecheng Jason Ma, Estelle Shen, Wanqiao Xu

Keywords: Risk-sensitive reinforcement learning, CVaR objective

TL;DR: We prove the first regret bounds for reinforcement learning under a general class of risk-sensitive objectives including the popular CVaR objective.

Abstract: In safety-critical applications of reinforcement learning such as healthcare and robotics, it is often desirable to optimize risk-sensitive objectives that account for tail outcomes rather than expected reward. We prove the first regret bounds for reinforcement learning under a general class of risk-sensitive objectives including the popular CVaR objective. Our theory is based on a novel characterization of the CVaR objective as well as a novel optimistic MDP construction.