When the Server Steps In: Calibrated Updates for Fair Federated Learning
Tianrun Yu, Kaixiang Zhao, Cheng Zhang, Anjun Gao, Yueyang Quan, Zhuqing Liu, Minghong Fang
TL;DR
This work tackles fairness in Federated Learning by introducing EquFL, a server-side debiasing method that blends a calibrated update with client updates to produce a fairer global model. The server constructs a synthetic dataset from early global models and derives a calibrated update by optimizing a chosen fairness metric on this synthetic data, then updates with $w^{t+1}=w^t-\eta_t(\gamma_t g_0^t+GAR(\cdot))$. Theoretical results guarantee convergence to the FedAvg optimum and reduced per-round fairness loss, while extensive experiments across six datasets and multiple metrics demonstrate substantial fairness gains with minimal accuracy loss and strong robustness to non-IID settings. The approach is aggregation-agnostic and fairness-metric agnostic, offering practical impact for deploying fair FL systems in diverse scenarios. Overall, EquFL provides a principled, scalable means to enhance fairness in FL without altering client training, with a clear path for extensions to multi-metric optimization and privacy-preserving deployments.
Abstract
Federated learning (FL) has emerged as a transformative distributed learning paradigm, enabling multiple clients to collaboratively train a global model under the coordination of a central server without sharing their raw training data. While FL offers notable advantages, it faces critical challenges in ensuring fairness across diverse demographic groups. To address these fairness concerns, various fairness-aware debiasing methods have been proposed. However, many of these approaches either require modifications to clients' training protocols or lack flexibility in their aggregation strategies. In this work, we address these limitations by introducing EquFL, a novel server-side debiasing method designed to mitigate bias in FL systems. EquFL operates by allowing the server to generate a single calibrated update after receiving model updates from the clients. This calibrated update is then integrated with the aggregated client updates to produce an adjusted global model that reduces bias. Theoretically, we establish that EquFL converges to the optimal global model achieved by FedAvg and effectively reduces fairness loss over training rounds. Empirically, we demonstrate that EquFL significantly mitigates bias within the system, showcasing its practical effectiveness.