Adversarial Bandits with Knapsacks

Author:

Immorlica Nicole1ORCID,Sankararaman Karthik2ORCID,Schapire Robert1ORCID,Slivkins Aleksandrs1ORCID

Affiliation:

1. Microsoft Research, Cambridge, MA, USA

2. Meta AI, Menlo Park, CA, USA

Abstract

We consider Bandits with Knapsacks (henceforth, BwK ), a general model for multi-armed bandits under supply/budget constraints. In particular, a bandit algorithm needs to solve a well-known knapsack problem : find an optimal packing of items into a limited-size knapsack. The BwK problem is a common generalization of numerous motivating examples, which range from dynamic pricing to repeated auctions to dynamic ad allocation to network routing and scheduling. While the prior work on BwK focused on the stochastic version, we pioneer the other extreme in which the outcomes can be chosen adversarially. This is a considerably harder problem, compared to both the stochastic version and the “classic” adversarial bandits, in that regret minimization is no longer feasible. Instead, the objective is to minimize the competitive ratio : the ratio of the benchmark reward to algorithm’s reward. We design an algorithm with competitive ratio O (log T ) relative to the best fixed distribution over actions, where T is the time horizon; we also prove a matching lower bound. The key conceptual contribution is a new perspective on the stochastic version of the problem. We suggest a new algorithm for the stochastic version, which builds on the framework of regret minimization in repeated games and admits a substantially simpler analysis compared to prior work. We then analyze this algorithm for the adversarial version, and use it as a subroutine to solve the latter. Our algorithm is the first “black-box reduction” from bandits to BwK: it takes an arbitrary bandit algorithm and uses it as a subroutine. We use this reduction to derive several extensions.

Publisher

Association for Computing Machinery (ACM)

Subject

Artificial Intelligence,Hardware and Architecture,Information Systems,Control and Systems Engineering,Software

Reference101 articles.

1. Jacob D. Abernethy and Jun-Kun Wang. 2017. On Frank-Wolfe and equilibrium computation. In Advances in Neural Information Processing Systems (NIPS). 6584–6593.

2. A reductions approach to fair classification;Agarwal Alekh;Fairness, Accountability, and Transparency in Machine Learning (FATML),2017

3. Alekh Agarwal, Miroslav Dudík, Satyen Kale, John Langford, and Robert E. Schapire. 2012. Contextual bandit learning with predictable rewards. In 15th Intl. Conf. on Artificial Intelligence and Statistics (AISTATS). 19–26.

4. Alekh Agarwal, Daniel Hsu, Satyen Kale, John Langford, Lihong Li, and Robert Schapire. 2014. Taming the monster: A fast and simple algorithm for contextual bandits. In 31st Intl. Conf. on Machine Learning (ICML).

5. Shipra Agrawal and Nikhil R. Devanur. 2014. Bandits with concave rewards and convex knapsacks. In 15th ACM Conf. on Economics and Computation (ACM-EC).

Cited by 5 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Supermodular Approximation of Norms and Applications;Proceedings of the 56th Annual ACM Symposium on Theory of Computing;2024-06-10

2. No-Regret Learning in Bilateral Trade via Global Budget Balance;Proceedings of the 56th Annual ACM Symposium on Theory of Computing;2024-06-10

3. Learning the Optimal Control for Evolving Systems with Converging Dynamics;Proceedings of the ACM on Measurement and Analysis of Computing Systems;2024-05-21

4. Liquid Welfare Guarantees for No-Regret Learning in Sequential Budgeted Auctions;Mathematics of Operations Research;2024-05-14

5. Technical Note—On Adaptivity in Nonstationary Stochastic Optimization with Bandit Feedback;Operations Research;2023-07-31

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3