Reinforcement Learning with Continuous Actions Under Unmeasured Confounding

• 成果类型: Article
• 署名作者: Li, Yuhan; Han, Eugene; Hu, Yifan; Zhou, Wenzhuo; Qi, Zhengling; Cui, Yifan; Zhu, Ruoqing
• 署名单位: University of Illinois System; University of Illinois Urbana-Champaign; University of Illinois System; University of Illinois Urbana-Champaign; University of California System; University of California Irvine; George Washington University; Zhejiang University; Zhejiang University
• 刊物名称: JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
• ISSN/ISSBN: 0162-1459; 1537-274X
• DOI: 10.1080/01621459.2025.2590175
• 发表日期: 2026-01-02
• 页码: 209-222
• 关键词: Causal Inference; Confounded POMDP; Policy Evaluation; policy optimization; Reinforcement Learning; safe

摘要

This article addresses the challenge of offline policy learning in continuous action spaces when unmeasured confounders are present. While most existing research focuses on policy evaluation within partially observable Markov decision processes (POMDPs) and assumes discrete action spaces, we advance this field by establishing a novel identification result to enable the nonparametric estimation of policy value for a given target policy under an infinite-horizon framework. Leveraging this identification, we develop a minimax estimator and introduce a policy-gradient-based algorithm to identify the in-class optimal policy that maximizes the estimated policy value. Furthermore, we provide theoretical results regarding the consistency, finite-sample error bound, and regret bound of the resulting optimal policy. Extensive simulations and a real-world application using the German Family Panel data demonstrate the effectiveness of our proposed methodology. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work.