Data-Free Distillation Improves Efficiency and Privacy in Federated Thorax Disease Analysis
This work addresses privacy and efficiency issues in federated learning for clinical thorax disease analysis, though it appears incremental as it builds on existing distillation methods.
The paper tackled the problem of thorax disease analysis in federated learning settings with privacy constraints by introducing FedKDF, a data-free distillation approach that aggregates knowledge without proxy datasets, resulting in improved efficiency and privacy preservation.
Thorax disease analysis in large-scale, multi-centre, and multi-scanner settings is often limited by strict privacy policies. Federated learning (FL) offers a potential solution, while traditional parameter-based FL can be limited by issues such as high communication costs, data leakage, and heterogeneity. Distillation-based FL can improve efficiency, but it relies on a proxy dataset, which is often impractical in clinical practice. To address these challenges, we introduce a data-free distillation-based FL approach FedKDF. In FedKDF, the server employs a lightweight generator to aggregate knowledge from different clients without requiring access to their private data or a proxy dataset. FedKDF combines the predictors from clients into a single, unified predictor, which is further optimized using the learned knowledge in the lightweight generator. Our empirical experiments demonstrate that FedKDF offers a robust solution for efficient, privacy-preserving federated thorax disease analysis.