FedCAR: Cross-client Adaptive Re-weighting for Generative Models in Federated Learning

TL;DR

This paper introduces FedCAR, a federated learning algorithm that adaptively re-weights client contributions based on distribution similarity of generated images, improving medical image generation without sharing raw data.

Contribution

FedCAR is the first to tailor adaptive re-weighting specifically for generative models in federated learning, enhancing performance and efficiency.

Findings

Outperforms centralized learning and traditional FL in medical image generation

Efficiently measures distribution similarity using generated images

Achieves superior results on chest X-ray datasets

Abstract

Generative models trained on multi-institutional datasets can provide an enriched understanding through diverse data distributions. However, training the models on medical images is often challenging due to hospitals' reluctance to share data for privacy reasons. Federated learning(FL) has emerged as a privacy-preserving solution for training distributed datasets across data centers by aggregating model weights from multiple clients instead of sharing raw data. Previous research has explored the adaptation of FL to generative models, yet effective aggregation algorithms specifically tailored for generative models remain unexplored. We hereby propose a novel algorithm aimed at improving the performance of generative models within FL. Our approach adaptively re-weights the contribution of each client, resulting in well-trained shared parameters. In each round, the server side measures the distribution distance between fake images generated by clients instead of directly comparing the Fr\'echet Inception Distance per client, thereby enhancing efficiency of the learning. Experimental results on three public chest X-ray datasets show superior performance in medical image generation, outperforming both centralized learning and conventional FL algorithms. Our code is available at https://github.com/danny0628/FedCAR.