Regulating Clients' Noise Adding in Federated Learning without Verification

Shu Hong, Lingjie Duan

In federated learning (FL), clients cooperatively train a global model without revealing their raw data but gradients or parameters, while the local information can still be disclo…