ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2310.19025
122
1
v1v2 (latest)

An Improved Relaxation for Oracle-Efficient Adversarial Contextual Bandits

29 October 2023
Kiarash Banihashem
Mohammadtaghi Hajiaghayi
Suho Shin
Max Springer
ArXiv (abs)PDFHTML
Abstract

We present an oracle-efficient relaxation for the adversarial contextual bandits problem, where the contexts are sequentially drawn i.i.d from a known distribution and the cost sequence is chosen by an online adversary. Our algorithm has a regret bound of O(T23(Klog⁡(∣Π∣))13)O(T^{\frac{2}{3}}(K\log(|\Pi|))^{\frac{1}{3}})O(T32​(Klog(∣Π∣))31​) and makes at most O(K)O(K)O(K) calls per round to an offline optimization oracle, where KKK denotes the number of actions, TTT denotes the number of rounds and Π\PiΠ denotes the set of policies. This is the first result to improve the prior best bound of O((TK)23(log⁡(∣Π∣))13)O((TK)^{\frac{2}{3}}(\log(|\Pi|))^{\frac{1}{3}})O((TK)32​(log(∣Π∣))31​) as obtained by Syrgkanis et al. at NeurIPS 2016, and the first to match the original bound of Langford and Zhang at NeurIPS 2007 which was obtained for the stochastic case.

View on arXiv
Comments on this paper