Federated Learning with Sparsified Model Perturbation: Improving Accuracy under Client-Level Differential Privacy

Open in new window