1 paper
Fatima Z. Abacha, Sin G. Teo, Yuanxiang Wu +2
Federated Learning remains highly susceptible to backdoor attacks--malicious clients inject targeted behaviours into the global model. Existing defenses suffer from substantial fal…