Skip to yearly menu bar Skip to main content


One-shot Empirical Privacy Estimation for Federated Learning

Galen Andrew · Peter Kairouz · Sewoong Oh · Alina Oprea · H. Brendan McMahan · Vinith Suriyakumar

Halle A 3
[ ] [ Visit Oral 7D ]
Fri 10 May 1 a.m. — 1:15 a.m. PDT


Privacy estimation techniques for differentially private (DP) algorithms are useful for comparing against analytical bounds, or to empirically measure privacy loss in settings where known analytical bounds are not tight. However, existing privacy auditing techniques usually make strong assumptions on the adversary (e.g., knowledge of intermediate model iterates or the training data distribution), are tailored to specific tasks, model architectures, or DP algorithm, and/or require retraining the model many times (typically on the order of thousands). These shortcomings make deploying such techniques at scale difficult in practice, especially in federated settings where model training can take days or weeks. In this work, we present a novel “one-shot” approach that can systematically address these challenges, allowing efficient auditing or estimation of the privacy loss of a model during the same, single training run used to fit model parameters, and without requiring any a priori knowledge about the model architecture, task, or DP algorithm. We show that our method provides provably correct estimates for the privacy loss under the Gaussian mechanism, and we demonstrate its performance on a well-established FL benchmark dataset under several adversarial threat models.

Chat is not available.