aiqbench analysis
This research advances empirical privacy auditing for differentially private machine learning by modeling canary signals as asymptotically Gaussian variables rather than binary thresholds. By leveraging this distributional perspective, the proposed framework derives tighter privacy lower bounds from a single training run of DP-SGD mechanisms. The work directly strengthens practical safeguards against membership inference attacks in modern ML pipelines without requiring multiple retraining runs.
Let's Ask Gauss: Improved One-Run Privacy Auditing
arXiv:2606.12733v2 Announce Type: replace Abstract: Privacy auditing provides an important safeguard by estimating the actual information leaked by a model, thus ensuring that theoretical privacy guarantees hold in practice. We study empirical privacy auditing for differentially private (DP) machine learning, focusing on efficient one-run methods for mechanisms such as DP-SGD. Prior one-run approaches threshold training examples or "canaries" into binary membership guesses, which discards useful information. We show that, in the white-box DP-SGD setting, canary-aligned signals naturally form a sequence of random variables whose normalized sum is asymptotically Gaussian. Leveraging this distributional perspective, we develop a DP-auditing framework that leads to tighter privacy lower bounds from a single training run.
Abstract:Privacy auditing provides an important safeguard by estimating the actual information leaked by a model, thus ensuring that theoretical privacy guarantees hold in practice. We study empirical privacy auditing for differentially private (DP) machine learning, focusing on efficient one-run methods for mechanisms such as DP-SGD. Prior one-run approaches threshold training examples or "canaries" into binary membership guesses, which discards useful information. We show that, in the white-box DP-SGD setting, canary-aligned signals naturally form a sequence of random variables whose normalized sum is asymptotically Gaussian. Leveraging this distributional perspective, we develop a DP-auditing framework that leads to tighter privacy lower bounds from a single training run.
Machine Learning (cs.LG)
arXiv:2606.12733 [cs.LG]
(or arXiv:2606.12733v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.12733 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 197 words
Related coverage
Let's Ask Gauss: Improved One-Run Privacy Auditing
arXiv · Oct 16, 2026
Optimizing Canaries for Privacy Auditing with Metagradient Descent
arXiv · Sep 23, 2026
Auditor-in-a-Box: Tools for Third-Party Auditing
LessWrong · Jul 28, 2026
[AINews] not much happened today
Latent Space · Sep 18, 2026