Enforcing group fairness in privacy-preserving Federated Learning

Chaomeng Chen, Zhenhong Zhou, Peng Tang, Longzhu He, Sen Su

Abstract: Highlights•We address group fairness in FL by focusing on diverse datasets’ distribution.•Introducing LSH-CADC for approximate global datasets, boosting data relevance.•Proposing AFA to enhance fairness in model aggregations with advanced weighting.•HybridAlpha merges into GFL, enhancing global fairness and privacy in FL.•GFL shows superior performance in fairness across varied FL settings and datasets.