Doubly robust off-policy evaluation with shrinkage
Yi Su, Maria Dimakopoulou, Akshay Krishnamurthy, Miroslav Dudik
Abstract
We propose a new framework for designing estimators for off-policy evaluation in contextual bandits. Our approach is based on the asymptotically optimal doubly robust estimator, but we shrink the importance weights to minimize a bound on the mean squared error, which results in a better bias-variance tradeoff in finite samples. We use this optimization-based framework to obtain three estimators: (a) a weight-clipping estimator, (b) a new weight-shrinkage estimator, and (c) the first shrinkage-based estimator for combinatorial action sets. Extensive experiments in both standard and combinatorial bandit benchmark problems show that our estimators are highly adaptive and typically outperform state-of-the-art methods.
BibTeX
@InProceedings{pmlr-v119-su20a,
title = {Doubly robust off-policy evaluation with shrinkage},
author = {Su, Yi and Dimakopoulou, Maria and Krishnamurthy, Akshay and Dudik, Miroslav},
booktitle = {Proceedings of the 37th International Conference on Machine Learning},
pages = {9167--9176},
year = {2020},
editor = {III, Hal Daumé and Singh, Aarti},
volume = {119},
series = {Proceedings of Machine Learning Research},
month = {13--18 Jul},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v119/su20a/su20a.pdf},
url = {https://proceedings.mlr.press/v119/su20a.html},
abstract = {We propose a new framework for designing estimators for off-policy evaluation in contextual bandits. Our approach is based on the asymptotically optimal doubly robust estimator, but we shrink the importance weights to minimize a bound on the mean squared error, which results in a better bias-variance tradeoff in finite samples. We use this optimization-based framework to obtain three estimators: (a) a weight-clipping estimator, (b) a new weight-shrinkage estimator, and (c) the first shrinkage-based estimator for combinatorial action sets. Extensive experiments in both standard and combinatorial bandit benchmark problems show that our estimators are highly adaptive and typically outperform state-of-the-art methods.}
}