Tight Auditing of Differential Privacy in MST and AIM
This work provides a rigorous auditing method for differential privacy in synthetic data, enabling tighter verification of privacy guarantees for practitioners using MST and AIM.
The paper introduces a Gaussian Differential Privacy (GDP)-based auditing framework for MST and AIM synthetic data generators, achieving the first tight audits in the strong-privacy regime. For (ε,δ)=(1,10^{-2}), the empirical privacy parameter μ_emp≈0.43 closely matches the theoretical μ=0.45, demonstrating a small theory-practice gap.
State-of-the-art Differentially Private (DP) synthetic data generators such as MST and AIM are widely used, yet tightly auditing their privacy guarantees remains challenging. We introduce a Gaussian Differential Privacy (GDP)-based auditing framework that measures privacy via the full false-positive/false-negative tradeoff. Applied to MST and AIM under worst-case settings, our method provides the first tight audits in the strong-privacy regime. For $(ε,δ)=(1,10^{-2})$, we obtain $μ_{emp}\approx0.43$ vs. implied $μ=0.45$, showing a small theory-practice gap. Our code is publicly available: https://github.com/sassoftware/dpmm.