An Online Learning Approach to a Multi-player N-armed Functional Bandit

O�Neill, Sam; Bagdasar, Ovidiu; Liotta, Antonio

Abstract

Congestion games possess the property of emitting at least one pure Nash equilibrium and have a rich history of practical use in transport modelling. In this paper we approach the problem of modelling equilibrium within congestion games using a decentralised multi-player probabilistic approach via stochastic bandit feedback. Restricting the strategies available to players under the assumption of bounded rationality, we explore an online multiplayer exponential weights algorithm for unweighted atomic routing games and compare this with a ϵ-greedy algorithm.

Citation

O’Neill, S., Bagdasar, O., & Liotta, A. (2020). An Online Learning Approach to a Multi-player N-armed Functional Bandit. In Numerical Computations: Theory and Algorithms (438-445). https://doi.org/10.1007/978-3-030-40616-5_41

• Presentation Conference Type: Conference Paper (Published)
• Conference Name: NUMTA: International Conference on Numerical Computations: Theory and Algorithms
• Start Date: Jun 15, 2019
• End Date: Jun 21, 2019
• Online Publication Date: Feb 14, 2020
• Publication Date: 2020
• Deposit Date: Apr 6, 2020
• Publisher: Springer
• Pages: 438-445
• Series Title: Lecture Notes in Computer Science
• Series Number: 11974
• Series ISSN: 0302-9743
• Book Title: Numerical Computations: Theory and Algorithms
• ISBN: 9783030406158
• DOI: https://doi.org/10.1007/978-3-030-40616-5_41
• Keywords: Congestion games, Online learning, Multi-armed bandit
• Public URL: http://researchrepository.napier.ac.uk/Output/2560144