Abstract: Federated learning (FL) enables the collaboration of multiple deep learning (DL) models to learn from decentralized data archives (i.e., clients) without accessing data on the clients.