Achieve Near-Optimal Individual Regret & Low Communications in Multi-Agent BanditsDownload PDF


22 Sept 2022, 12:38 (modified: 11 Nov 2022, 14:08)ICLR 2023 Conference Blind SubmissionReaders: Everyone
Keywords: Multi-agent multi-armed bandits, individual regret, communication
TL;DR: A near-optimal algorithm for both individual and group regrets and only requiring O(\log (\log T)) communication times
Abstract: Cooperative multi-agent multi-armed bandits (\CMAB) study how distributed agents cooperatively play the same multi-armed bandit game. Most existing \CMAB works focused on maximizing the group performance of all agents---the accumulation of all agents' individual performance (i.e., individual reward). However, in many applications, the performance of the system is more sensitive to the ``bad'' agent---the agent with the worst individual performance. For example, in a drone swarm, a ``bad'' agent may crash into other drones and severely degrade the system performance. In that case, the key of the learning algorithm design is to coordinate computational and communicational resources among agents so to optimize the individual learning performance of the ``bad'' agent. In \CMAB, maximizing the group performance is equivalent to minimizing the group regret of all agents, and minimizing the individual performance can be measured by minimizing the maximum (worst) individual regret among agents. Minimizing the maximum individual regret was largely ignored in prior literature, and currently, there is little work on how to minimize this objective with a low communication overhead. In this paper, we propose a near-optimal algorithm on both individual and group regrets, in addition, we also propose a novel communication module in the algorithm, which only needs \(O(\log (\log T))\) communication times where \(T\) is the number of decision rounds. We also conduct simulations to illustrate the advantage of our algorithm by comparing it to other known baselines.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics
Submission Guidelines: Yes
Please Choose The Closest Area That Your Submission Falls Into: Reinforcement Learning (eg, decision and control, planning, hierarchical RL, robotics)
15 Replies