Purifying Gradients Protects Decentralized Federated Learning Models Against Data Poisoning Attacks
Zhejiang University researchers developed a gradient purification defense that eliminates malicious updates while retaining beneficial information in decentralized federated learning ...
