Result Details

Cross-Validated Off-Policy Evaluation

ČIEF, M.; KVETON, B.; KOMPAN, M. Cross-Validated Off-Policy Evaluation. In Proceedings of the AAAI Conference on Artificial Intelligence. Pennsylvania: 2025. p. 16073-16081. ISBN: 978-1-57735-897-8.
Type
conference paper
Language
English
Authors
Čief Matej, Ing., Ph.D., DCGM (FIT)
Kveton Branislav
Kompan Michal, doc. Ing., PhD., DCGM (FIT)
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.

Keywords

off-policy evaluation, cross-validation, estimator selection, hyper-parameter tuning

URL
Published
2025
Pages
16073–16081
Proceedings
Proceedings of the AAAI Conference on Artificial Intelligence
Conference
The 39th Annual AAAI Conference on Artificial Intelligence
ISBN
978-1-57735-897-8
Place
Pennsylvania
DOI
10.1609/aaai.v39i15.33765
UT WoS
001477532300090
BibTeX 
@inproceedings{BUT197085,
  author="Matej {Čief} and  {} and Michal {Kompan}",
  title="Cross-Validated Off-Policy Evaluation",
  booktitle="Proceedings of the AAAI Conference on Artificial Intelligence",
  year="2025",
  pages="16073--16081",
  address="Pennsylvania",
  doi="10.1609/aaai.v39i15.33765",
  isbn="978-1-57735-897-8",
  url="https://ojs.aaai.org/index.php/AAAI/article/view/33765"
}
Files
Departments
Department of Computer Graphics and Multimedia (DCGM)
Back to top