Efficient Privacy Auditing in Federated Learning

We design a novel efficient membership inference attack to audit privacy risks in federated learning. Our approach involves computing the slope of specific model performance metrics (e.g., model's output and its loss) across FL rounds to differentiate members from non-members. Since these metrics are automatically computed during the FL process, our solution imposes negligible overhead and can be seamlessly integrated without disrupting training. We validate the effectiveness and superiority of our method over prior work across a wide range of FL settings and real-world datasets.