Defending Non-Bayesian Learning against Adversarial Attacks

TL;DR

This paper investigates how adversarial Byzantine agents impact non-Bayesian learning in multi-agent networks and proposes two new learning rules to mitigate these effects.

Contribution

It introduces two novel learning rules designed to enhance robustness of non-Bayesian learning against Byzantine faults in multi-agent systems.

Findings

Proposed learning rules improve resilience to Byzantine attacks

Demonstrated effectiveness through theoretical analysis

Validated robustness via simulations

Abstract

This paper addresses the problem of non-Bayesian learning over multi-agent networks, where agents repeatedly collect partially informative observations about an unknown state of the world, and try to collaboratively learn the true state. We focus on the impact of the adversarial agents on the performance of consensus-based non-Bayesian learning, where non-faulty agents combine local learning updates with consensus primitives. In particular, we consider the scenario where an unknown subset of agents suffer Byzantine faults -- agents suffering Byzantine faults behave arbitrarily. Two different learning rules are proposed.