Personalized Federated Learning via Heterogeneous Modular Networks

TL;DR

This paper introduces FedMN, a personalized federated learning method that adaptively assembles heterogeneous neural networks for clients, improving personalization and efficiency under data divergence and privacy constraints.

Contribution

FedMN proposes a novel modular network approach with a hypernetwork for personalized model assembly in federated learning, reducing communication costs.

Findings

FedMN outperforms baseline methods in accuracy and personalization.

The approach reduces communication overhead in federated settings.

Experimental results validate the effectiveness and efficiency of FedMN.

Abstract

Personalized Federated Learning (PFL) which collaboratively trains a federated model while considering local clients under privacy constraints has attracted much attention. Despite its popularity, it has been observed that existing PFL approaches result in sub-optimal solutions when the joint distribution among local clients diverges. To address this issue, we present Federated Modular Network (FedMN), a novel PFL approach that adaptively selects sub-modules from a module pool to assemble heterogeneous neural architectures for different clients. FedMN adopts a light-weighted routing hypernetwork to model the joint distribution on each client and produce the personalized selection of the module blocks for each client. To reduce the communication burden in existing FL, we develop an efficient way to interact between the clients and the server. We conduct extensive experiments on the real-world test beds and the results show both the effectiveness and efficiency of the proposed FedMN over the baselines.