|国家预印本平台
首页|Percentile Criterion Optimization in Offline Reinforcement Learning

Percentile Criterion Optimization in Offline Reinforcement Learning

Percentile Criterion Optimization in Offline Reinforcement Learning

来源:Arxiv_logoArxiv
英文摘要

In reinforcement learning, robust policies for high-stakes decision-making problems with limited data are usually computed by optimizing the \emph{percentile criterion}. The percentile criterion is approximately solved by constructing an \emph{ambiguity set} that contains the true model with high probability and optimizing the policy for the worst model in the set. Since the percentile criterion is non-convex, constructing ambiguity sets is often challenging. Existing work uses \emph{Bayesian credible regions} as ambiguity sets, but they are often unnecessarily large and result in learning overly conservative policies. To overcome these shortcomings, we propose a novel Value-at-Risk based dynamic programming algorithm to optimize the percentile criterion without explicitly constructing any ambiguity sets. Our theoretical and empirical results show that our algorithm implicitly constructs much smaller ambiguity sets and learns less conservative robust policies.

Yair Zick、Cyrus Cousins、Elita A. Lobo、Marek Petrik

计算技术、计算机技术

Yair Zick,Cyrus Cousins,Elita A. Lobo,Marek Petrik.Percentile Criterion Optimization in Offline Reinforcement Learning[EB/OL].(2024-04-07)[2025-07-22].https://arxiv.org/abs/2404.05055.点此复制

评论