Communication-Efficient Federated AUC Maximization with Cyclic Client Participation
Umesh Vangapally, Wenhan Wu, Chen Chen, Zhishuai Guo
TL;DR
This work tackles federated AUC maximization under realistic cyclic client participation (CyCP) by developing two families of algorithms: a stagewise minimax method for squared surrogate AUC and an active–passive gradient framework for general pairwise AUC losses. Under the Polyak–Łojasiewicz (PL) condition, the authors establish strong convergence guarantees and favorable communication complexities, notably achieving $\widetilde{O}(1/\epsilon^{1/2})$ communication with linear speedup in simultaneous participants $M$ for minimax, and $\widetilde{O}(1/\epsilon^{1/2})$ (under PL) for pairwise objectives. The methods are validated on diverse benchmarks (image classification, medical imaging, fraud detection), showing superior efficiency and robustness to non-IID data, class imbalance, and label noise. By addressing the non-decomposable AUC objective under cyclic participation, the paper provides practical tools for privacy-preserving, communication-efficient federated learning in real-world deployments.
Abstract
Federated AUC maximization is a powerful approach for learning from imbalanced data in federated learning (FL). However, existing methods typically assume full client availability, which is rarely practical. In real-world FL systems, clients often participate in a cyclic manner: joining training according to a fixed, repeating schedule. This setting poses unique optimization challenges for the non-decomposable AUC objective. This paper addresses these challenges by developing and analyzing communication-efficient algorithms for federated AUC maximization under cyclic client participation. We investigate two key settings: First, we study AUC maximization with a squared surrogate loss, which reformulates the problem as a nonconvex-strongly-concave minimax optimization. By leveraging the Polyak-Łojasiewicz (PL) condition, we establish a state-of-the-art communication complexity of $\widetilde{O}(1/ε^{1/2})$ and iteration complexity of $\widetilde{O}(1/ε)$. Second, we consider general pairwise AUC losses. We establish a communication complexity of $O(1/ε^3)$ and an iteration complexity of $O(1/ε^4)$. Further, under the PL condition, these bounds improve to communication complexity of $\widetilde{O}(1/ε^{1/2})$ and iteration complexity of $\widetilde{O}(1/ε)$. Extensive experiments on benchmark tasks in image classification, medical imaging, and fraud detection demonstrate the superior efficiency and effectiveness of our proposed methods.