Federated learning (FL) is emerging as a new paradigm to train machine learning (ML) models in distributed systems. Rather than sharing and disclosing the training data set with the server, the model parameters (e.g., neural networks’ weights and biases) are optimized collectively by large popula…