Abstract: Federated learning, as a privacy preserving distributed machine learning paradigm, enables geographically dispersed clients to collaboratively train a shared model without exchanging raw ...