BaFFLe: Backdoor detection via Feedback-based Federated Learning

no code implementations • 4 Nov 2020 • Sebastien Andreina, Giorgia Azzurra Marson, Helen Möllering, Ghassan Karame

In this paper, we propose Backdoor detection via Feedback-based Federated Learning (BAFFLE), a novel defense to secure FL against backdoor attacks.

Federated Learning Model Poisoning