Beyond UCB: Optimal and Efficient Contextual Bandits with Regression Oracles
Author(s)
Foster, Dylan J; Rakhlin, Alexander
DownloadPublished version (527.1Kb)
Publisher Policy
Publisher Policy
Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use.
Terms of use
Metadata
Show full item recordAbstract
A fundamental challenge in contextual bandits is to develop flexible, general-purpose algorithms with computational requirements no worse than classical supervised learning tasks such as classification and regression. Algorithms based on regression have shown promising empirical success, but theoretical guarantees have remained elusive except in special cases. We provide the first universal and optimal reduction from contextual bandits to online regression. We show how to transform any oracle for online regression with a given value function class into an algorithm for contextual bandits with the induced policy class, with no overhead in runtime or memory requirements. We characterize the minimax rates for contextual bandits with general, potentially nonparametric function classes, and show that our algorithm is minimax optimal whenever the oracle obtains the optimal rate for regression. Compared to previous results, our algorithm requires no distributional assumptions beyond realizability, and works even when contexts are chosen adversarially.
Date issued
2020Department
Statistics and Data Science Center (Massachusetts Institute of Technology); Massachusetts Institute of Technology. Institute for Data, Systems, and Society; Massachusetts Institute of Technology. Laboratory for Information and Decision Systems; Massachusetts Institute of Technology. Department of Brain and Cognitive SciencesJournal
INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 119
Citation
Foster, Dylan J and Rakhlin, Alexander. 2020. "Beyond UCB: Optimal and Efficient Contextual Bandits with Regression Oracles." INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 119, 119.
Version: Final published version