On the Effect of Defection in Federated Learning and How to Prevent It

TL;DR — We study the effect of devices permanently dropping out of federated optimization and provide a new algorithm to provably avoid defections.

Abstract

Federated learning is a machine learning protocol that enables a large population of agents to collaborate. These agents communicate over multiple rounds to produce a single, consensus model. Despite this collaborative framework, there are instances where agents may choose to defect permanently—essentially withdrawing from the collaboration—if they are content with their instantaneous model in that round. This work demonstrates the detrimental impact such defections can have on the final model's robustness and ability to generalize. We also show that current federated optimization algorithms fall short in disincentivizing these harmful defections. To address this, we introduce a novel optimization algorithm with theoretical guarantees to prevent defections while ensuring asymptotic convergence to an effective solution for all participating agents. We also provide numerical experiments to corroborate our findings and demonstrate the effectiveness of our algorithm.