Personalized federated learning based on feature fusion

TL;DR

This paper introduces pFedPM, a personalized federated learning method that uses feature fusion and a relation network to improve classification performance under data heterogeneity while reducing communication costs.

Contribution

The paper proposes a novel personalized federated learning approach that replaces gradient uploading with feature uploading and incorporates a relation network for improved classification.

Findings

Outperforms recent FL methods on MNIST, FEMNIST, and CIFAR10 datasets.

Reduces communication costs compared to traditional methods.

Effectively handles label distribution skew in federated learning.

Abstract

Federated learning enables distributed clients to collaborate on training while storing their data locally to protect client privacy. However, due to the heterogeneity of data, models, and devices, the final global model may need to perform better for tasks on each client. Communication bottlenecks, data heterogeneity, and model heterogeneity have been common challenges in federated learning. In this work, we considered a label distribution skew problem, a type of data heterogeneity easily overlooked. In the context of classification, we propose a personalized federated learning approach called pFedPM. In our process, we replace traditional gradient uploading with feature uploading, which helps reduce communication costs and allows for heterogeneous client models. These feature representations play a role in preserving privacy to some extent. We use a hyperparameter $a$ to mix local and global features, which enables us to control the degree of personalization. We also introduced a relation network as an additional decision layer, which provides a non-linear learnable classifier to predict labels. Experimental results show that, with an appropriate setting of $a$, our scheme outperforms several recent FL methods on MNIST, FEMNIST, and CRIFAR10 datasets and achieves fewer communications.