FedCAR: Cross-client Adaptive Re-weighting for Generative Models in Federated Learning
This work addresses privacy concerns in medical imaging by improving generative model training in federated settings, though it is incremental as it builds on existing FL methods with a tailored aggregation algorithm.
The paper tackled the challenge of training generative models on distributed medical image datasets without sharing raw data by proposing FedCAR, a federated learning algorithm that adaptively re-weights client contributions based on distribution distances between generated images, resulting in superior performance over centralized and conventional FL methods on chest X-ray datasets.
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 Fréchet 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.