Doubly adversarial federated bandits

Yi, Jialin and Vojnović, Milan ORCID: 0000-0003-1382-022X (2023) Doubly adversarial federated bandits. Proceedings of Machine Learning Research. 39951 - 39967. ISSN 1938-7228

Text (Doubly adversarial federated bandits) - Published Version
Download (1MB)

Abstract

We study a new non-stochastic federated multiarmed bandit problem with multiple agents collaborating via a communication network. The losses of the arms are assigned by an oblivious adversary that specifies the loss of each arm not only for each time step but also for each agent, which we call “doubly adversarial”. In this setting, different agents may choose the same arm in the same time step but observe different feedback. The goal of each agent is to find a globally best arm in hindsight that has the lowest cumulative loss averaged over all agents, which necessities the communication among agents. We provide regret lower bounds for any federated bandit algorithm under different settings, when agents have access to full-information feedback, or the bandit feedback. For the bandit feedback setting, we propose a near-optimal federated bandit algorithm called FEDEXP3. Our algorithm gives a positive answer to an open question proposed in (Cesa-Bianchi et al., 2016): FEDEXP3 can guarantee a sub-linear regret without exchanging sequences of selected arm identities or loss sequences among agents. We also provide numerical evaluations of our algorithm to validate our theoretical results and demonstrate its effectiveness on synthetic and real-world datasets.

Item Type:	Article
Official URL:	https://dl.acm.org/doi/proceedings/10.5555/3618408
Additional Information:	© 2023 The Authors
Divisions:	Statistics
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Date Deposited:	28 May 2024 10:21
Last Modified:	21 Nov 2024 21:54
URI:	http://eprints.lse.ac.uk/id/eprint/123667

Actions (login required)

View Item

Download Statistics

Downloads

Downloads per month over past year

View more statistics