Bias in Federated Learning: A Comprehensive Survey

Abstract

Federated Learning (FL) enables collaborative model training over multiple clients’ data, without sharing these data for better privacy. Addressing bias in FL remains a challenge. In this article, we first present a taxonomy of FL bias, presenting the causes and the different types of FL bias, namely demographic bias, performance-related bias, and contribution-related bias. We then categorize FL bias mitigation, in terms of used methods and provided guarantees, before providing a comprehensive and comparative analysis of existing works. Finally, we highlight key challenges and open research directions, including the impact of FL bias mitigation on model utility, privacy, and robustness.

Topics & Concepts