Publication Details

Cross-Validated Off-Policy Evaluation

ČIEF Matej and KOMPAN Michal. Cross-Validated Off-Policy Evaluation. In: Proceedings of the AAAI Conference on Artificial Intelligence. Pennsylvania, 2025, pp. 16073-16081. ISBN 978-1-57735-897-8. Available from: https://ojs.aaai.org/index.php/AAAI/article/view/33765
Type
conference paper
Language
english
Authors
Čief Matej, Ing. (DCGM FIT BUT)
Kompan Michal, doc. Ing., Ph.D. (DCGM FIT BUT)
URL
Abstract

We study estimator selection and hyper-parameter tuning in off-policy evaluation. Although cross-validation is the most popular method for model selection in supervised learning, off-policy evaluation relies mostly on theory, which provides only limited guidance to practitioners. We show how to use cross-validation for off-policy evaluation. This challenges a popular belief that cross-validation in off-policy evaluation is not feasible. We evaluate our method empirically and show that it addresses a variety of use cases.

Published
2025
Pages
16073-16081
Proceedings
Proceedings of the AAAI Conference on Artificial Intelligence
Conference
The 39th Annual AAAI Conference on Artificial Intelligence, Philadelphia, US
ISBN
978-1-57735-897-8
Place
Pennsylvania, US
DOI
10.1609/aaai.v39i15.33765
BibTeX 
@INPROCEEDINGS{FITPUB13317,
   author = "Matej \v{C}ief and Michal Kompan",
   title = "Cross-Validated Off-Policy Evaluation",
   pages = "16073--16081",
   booktitle = "Proceedings of the AAAI Conference on Artificial Intelligence",
   year = 2025,
   location = "Pennsylvania, US",
   ISBN = "978-1-57735-897-8",
   doi = "10.1609/aaai.v39i15.33765",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/13317"
}
Back to top