Machine learningML & AI Methods

FilFL: Client Filtering for Optimized Client Participation in Federated Learning

Client filtering, a new method to enhance model generalization and client participation in federated learning, has been proposed, leading to better learning efficiency, quicker convergence, and up to 10% increased test accuracy.