Personalized Multi-tier Federated Learning
This addresses statistical heterogeneity in federated learning for devices with team structures, offering incremental improvements in communication efficiency and customization.
The paper tackles the challenge of personalized federated learning by introducing a multi-tier architecture (PerMFL) to optimize local models for devices with known team structures, achieving linear convergence for convex problems and sub-linear for non-convex ones, and outperforming state-of-the-art methods in experiments.
The key challenge of personalized federated learning (PerFL) is to capture the statistical heterogeneity properties of data with inexpensive communications and gain customized performance for participating devices. To address these, we introduced personalized federated learning in multi-tier architecture (PerMFL) to obtain optimized and personalized local models when there are known team structures across devices. We provide theoretical guarantees of PerMFL, which offers linear convergence rates for smooth strongly convex problems and sub-linear convergence rates for smooth non-convex problems. We conduct numerical experiments demonstrating the robust empirical performance of PerMFL, outperforming the state-of-the-art in multiple personalized federated learning tasks.