Asymptotic Optimality for Decentralised Bandits

Abstract We consider a large number of agents collaborating on multi-armed bandit problem with arms. The goal is to minimise the regret each agent in communication-constrained setting. present decentralised algorithm which builds upon and improves Gossip-Insert-Eliminate method Chawla et al. (International conference artificial intelligence statistics, pp 3471–3481, 2020). provide theoretical analysis incurred shows that our asymptotically optimal. In fact, guarantee matches optimal rate achievable full communication Finally, we empirical results support conclusions.