FedMIA: An Effective Membership Inference Attack Exploiting "All for One" Principle in Federated Learning

Gongxi Zhu, Donghao Li, Hanlin Gu +3

Federated Learning (FL) is a promising approach for training machine learning models on decentralized data while preserving privacy. However, privacy risks, particularly Membership…