Provably Efficient Iterated CVaR Reinforcement Learning with Function Approximation and Human Feedback
Yu Chen, Yihan Du, Pihe Hu, Siwei Wang, Desheng Wu, Longbo Huang
Abstract
Risk-sensitive reinforcement learning (RL) aims to optimize policies that balance the expected reward and risk. In this paper, we present a novel risk-sensitive RL framework that employs an Iterated Conditional Value-at-Risk (CVaR) objective under both linear and general function approximations, enriched by human feedback. These new formulations provide a principled way to guarantee safety in each decision making step throughout the control process. Moreover, integrating human feedback into risk-sensitive RL framework bridges the gap between algorithmic decision-making and human participation, allowing us to also guarantee safety for human-in-the-loop systems. We propose provably sample-efficient algorithms for this Iterated CVaR RL and provide rigorous theoretical analysis. Furthermore, we establish a matching lower bound to corroborate the optimality of our algorithms in a linear context.
BibTeX
@inproceedings{
chen2024provably,
title={Provably Efficient Iterated {CV}aR Reinforcement Learning with Function Approximation and Human Feedback},
author={Yu Chen and Yihan Du and Pihe Hu and Siwei Wang and Desheng Wu and Longbo Huang},
booktitle={The Twelfth International Conference on Learning Representations},
year={2024},
url={https://openreview.net/forum?id=vW1SkPl4kp}
}