LiD-FL: Towards List-Decodable Federated Learning

TL;DR

This paper introduces a list-decodable federated learning framework that tolerates a majority of malicious participants, providing convergence guarantees and demonstrating robustness through experiments on image classification tasks.

Contribution

It presents a novel list-decodable federated learning algorithm that works without strict honest worker fraction constraints, extending Byzantine robustness.

Findings

The algorithm converges under certain loss function assumptions.

It withstands malicious majority in various attack scenarios.

Experimental results confirm robustness on image classification tasks.

Abstract

Federated learning is often used in environments with many unverified participants. Therefore, federated learning under adversarial attacks receives significant attention. This paper proposes an algorithmic framework for list-decodable federated learning, where a central server maintains a list of models, with at least one guaranteed to perform well. The framework has no strict restriction on the fraction of honest workers, extending the applicability of Byzantine federated learning to the scenario with more than half adversaries. Under proper assumptions on the loss function, we prove a convergence theorem for our method. Experimental results, including image classification tasks with both convex and non-convex losses, demonstrate that the proposed algorithm can withstand the malicious majority under various attacks.