Sparse Federated Learning with Hierarchical Personalized Models

TL;DR

This paper introduces sFedHP, a hierarchical personalized federated learning algorithm that enhances model performance on diverse data, reduces communication costs through sparsity, and maintains fast convergence.

Contribution

The paper presents a novel hierarchical personalized FL method with a sparse constraint, improving performance on non-i.i.d. data and reducing communication overhead.

Findings

sFedHP outperforms FedAvg, HierFAVG, and other personalized FL methods.

The sparse constraint significantly reduces communication costs.

Convergence rate remains state-of-the-art with only minor slowdown due to sparsity.

Abstract

Federated learning (FL) can achieve privacy-safe and reliable collaborative training without collecting users' private data. Its excellent privacy security potential promotes a wide range of FL applications in Internet-of-Things (IoT), wireless networks, mobile devices, autonomous vehicles, and cloud medical treatment. However, the FL method suffers from poor model performance on non-i.i.d. data and excessive traffic volume. To this end, we propose a personalized FL algorithm using a hierarchical proximal mapping based on the moreau envelop, named sparse federated learning with hierarchical personalized models (sFedHP), which significantly improves the global model performance facing diverse data. A continuously differentiable approximated L1-norm is also used as the sparse constraint to reduce the communication cost. Convergence analysis shows that sFedHP's convergence rate is state-of-the-art with linear speedup and the sparse constraint only reduces the convergence rate to a small extent while significantly reducing the communication cost. Experimentally, we demonstrate the benefits of sFedHP compared with the FedAvg, HierFAVG (hierarchical FedAvg), and personalized FL methods based on local customization, including FedAMP, FedProx, Per-FedAvg, pFedMe, and pFedGP.