Exploit Gradient Skewness to Circumvent Byzantine Defenses for Federated Learning

Open in new window