Cooperative Stochastic Multi-agent Multi-armed Bandits Robust to Adversarial Corruptions

Abstract

We study the problem of stochastic bandits with adversarial corruptions in the cooperative multi-agent setting, where V agents interact with a common K-armed bandit problem, and each pair of agents can communicate with each other to expedite the learning process. In the problem, the rewards are independently sampled from distributions across all agents and rounds, but they may be corrupted by an adversary. Our goal is to minimize both the overall regret and communication cost across all agents. We first show that an additive term of corruption is unavoidable for any algorithm in this problem. Then, we propose a new algorithm that is agnostic to the level of corruption. Our algorithm not only achieves near-optimal regret in the stochastic setting, but also obtains a regret with an additive term of corruption in the corrupted setting, while maintaining efficient communication. The algorithm is also applicable for the single-agent corruption problem, and achieves a high probability regret that removes the multiplicative dependence of K on corruption level. Our result of the single-agent case resolves an open question from Gupta et al. [2019].Preprint. Under review.

Robust Multi-Agent Bandits Over Undirected Graphs

“…Most of their paper involves a different communication model where the agents/clients collaborate via a central server; Section 6 studies a "peer-to-peer" model which is closer to ours but requires additional assumptions on the number of malicious neighbors. A different line of work considers the case where an adversary can corrupt the observed rewards (see, e.g., [11,12,25,26,29,33,39,40,43], and the references therein), which is distinct from the role that malicious agents play in our setting.…”

An Attackability Perspective on No-Sensing Adversarial Multi-player Multi-armed Bandits

“…IV). Other forms of attacks are also studied [28]- [31], including the "weak attack" model [32]- [34] where attacks are performed before observing actions. Note that the attackers in all these works have no desire to explore the environment, while the reward-teaching server has to actively learn the global model.…”