Bias-Robust Bayesian Optimization via Dueling Bandits

Johannes Kirschner, Andreas Krause
Proceedings of the 38th International Conference on Machine Learning, PMLR 139:5595-5605, 2021.

Abstract

We consider Bayesian optimization in settings where observations can be adversarially biased, for example by an uncontrolled hidden confounder. Our first contribution is a reduction of the confounded setting to the dueling bandit model. Then we propose a novel approach for dueling bandits based on information-directed sampling (IDS). Thereby, we obtain the first efficient kernelized algorithm for dueling bandits that comes with cumulative regret guarantees. Our analysis further generalizes a previously proposed semi-parametric linear bandit model to non-linear reward functions, and uncovers interesting links to doubly-robust estimation.

Cite this Paper

BibTeX
@InProceedings{pmlr-v139-kirschner21a, title = {Bias-Robust Bayesian Optimization via Dueling Bandits}, author = {Kirschner, Johannes and Krause, Andreas}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, pages = {5595--5605}, year = {2021}, editor = {Meila, Marina and Zhang, Tong}, volume = {139}, series = {Proceedings of Machine Learning Research}, month = {18--24 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v139/kirschner21a/kirschner21a.pdf}, url = {https://proceedings.mlr.press/v139/kirschner21a.html}, abstract = {We consider Bayesian optimization in settings where observations can be adversarially biased, for example by an uncontrolled hidden confounder. Our first contribution is a reduction of the confounded setting to the dueling bandit model. Then we propose a novel approach for dueling bandits based on information-directed sampling (IDS). Thereby, we obtain the first efficient kernelized algorithm for dueling bandits that comes with cumulative regret guarantees. Our analysis further generalizes a previously proposed semi-parametric linear bandit model to non-linear reward functions, and uncovers interesting links to doubly-robust estimation.} }
Endnote
%0 Conference Paper %T Bias-Robust Bayesian Optimization via Dueling Bandits %A Johannes Kirschner %A Andreas Krause %B Proceedings of the 38th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2021 %E Marina Meila %E Tong Zhang %F pmlr-v139-kirschner21a %I PMLR %P 5595--5605 %U https://proceedings.mlr.press/v139/kirschner21a.html %V 139 %X We consider Bayesian optimization in settings where observations can be adversarially biased, for example by an uncontrolled hidden confounder. Our first contribution is a reduction of the confounded setting to the dueling bandit model. Then we propose a novel approach for dueling bandits based on information-directed sampling (IDS). Thereby, we obtain the first efficient kernelized algorithm for dueling bandits that comes with cumulative regret guarantees. Our analysis further generalizes a previously proposed semi-parametric linear bandit model to non-linear reward functions, and uncovers interesting links to doubly-robust estimation.
APA
Kirschner, J. & Krause, A.. (2021). Bias-Robust Bayesian Optimization via Dueling Bandits. Proceedings of the 38th International Conference on Machine Learning, in Proceedings of Machine Learning Research 139:5595-5605 Available from https://proceedings.mlr.press/v139/kirschner21a.html.

Related Material