Abstract: Federated Learning (FL) is to collaboratively train a global model among distributed clients by iteratively aggregating their local updates without sharing their raw data, whereby the global ...