We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Machine Learning

Title: Towards Optimal Algorithms for Multi-Player Bandits without Collision Sensing Information

Abstract: We propose a novel algorithm for multi-player multi-armed bandits without collision sensing information. Our algorithm circumvents two problems shared by all state-of-the-art algorithms: it does not need as an input a lower bound on the minimal expected reward of an arm, and its performance does not scale inversely proportionally to the minimal expected reward. We prove a theoretical regret upper bound to justify these claims. We complement our theoretical results with numerical experiments, showing that the proposed algorithm outperforms state-of-the-art in practice as well.
Comments: 24 pages, COLT 2022
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Cite as: arXiv:2103.13059 [stat.ML]
  (or arXiv:2103.13059v2 [stat.ML] for this version)

Submission history

From: Richard Combes [view email]
[v1] Wed, 24 Mar 2021 10:14:16 GMT (250kb,D)
[v2] Mon, 6 Jun 2022 07:31:32 GMT (519kb,D)

Link back to: arXiv, form interface, contact.