Detecting Backdoor Attacks in Federated Learning via Direction Alignment Inspection

Open in new window