Saved in:
Bibliographic Details
Main Authors: Wan, Yilong, Li, Yuqiang, Wu, Xianyi
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.16236
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908460600262656
author Wan, Yilong
Li, Yuqiang
Wu, Xianyi
author_facet Wan, Yilong
Li, Yuqiang
Wu, Xianyi
contents This paper investigates off-policy evaluation in contextual bandits, aiming to quantify the performance of a target policy using data collected under a different and potentially unknown behavior policy. Recently, methods based on conformal prediction have been developed to construct reliable prediction intervals that guarantee marginal coverage in finite samples, making them particularly suited for safety-critical applications. To further achieve coverage conditional on a given offline data set, we propose a novel algorithm that constructs probably approximately correct prediction intervals. Our method builds upon a PAC-valid conformal prediction framework, and we strengthen its theoretical guarantees by establishing PAC-type bounds on coverage. We analyze both finite-sample and asymptotic properties of the proposed method, and compare its empirical performance with existing methods in simulations.
format Preprint
id arxiv_https___arxiv_org_abs_2507_16236
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PAC Off-Policy Prediction of Contextual Bandits
Wan, Yilong
Li, Yuqiang
Wu, Xianyi
Machine Learning
This paper investigates off-policy evaluation in contextual bandits, aiming to quantify the performance of a target policy using data collected under a different and potentially unknown behavior policy. Recently, methods based on conformal prediction have been developed to construct reliable prediction intervals that guarantee marginal coverage in finite samples, making them particularly suited for safety-critical applications. To further achieve coverage conditional on a given offline data set, we propose a novel algorithm that constructs probably approximately correct prediction intervals. Our method builds upon a PAC-valid conformal prediction framework, and we strengthen its theoretical guarantees by establishing PAC-type bounds on coverage. We analyze both finite-sample and asymptotic properties of the proposed method, and compare its empirical performance with existing methods in simulations.
title PAC Off-Policy Prediction of Contextual Bandits
topic Machine Learning
url https://arxiv.org/abs/2507.16236