The Hidden Adversarial Vulnerabilities of Medical Federated Learning

Erfan Darzi, Florian Dubost, Nanna. M. Sijtsema +1

In this paper, we delve into the susceptibility of federated medical image analysis systems to adversarial attacks. Our analysis uncovers a novel exploitation avenue: using gradien…