Federated learning, where the distribution of distributed data is unknown, is more difficult and costly to train a central model with than traditional machine learning. In this study, we propose ...