Tight Privacy Audit in One Run

2025-09-10

AI Summary

Auditing differential privacy (DP) guarantees in a single run remains challenging, especially for complex formulations such as (ε,δ)-DP, where existing methods lack coverage and theoretical tightness. Method: We propose the first provably tight single-run auditing framework, grounded in f-DP theory and integrating statistical hypothesis testing with convex optimization to enable one-shot, precise, and efficient privacy budget estimation for arbitrary DP mechanisms. Contribution/Results: Theoretically, we establish the first tight lower bound on auditability. Technically, we overcome the fundamental limitations of prior approaches in (ε,δ)-DP settings, where they provably fail. Empirically, our framework consistently outperforms state-of-the-art methods across diverse DP algorithms—including Laplace, Gaussian, and randomized response—demonstrating both tightness and practical efficiency. Moreover, it corrects previously held misconceptions regarding parameter sensitivity in DP auditing.

Technology Category

• Machine Learning: Privacy
• Reasoning under Uncertainty: Stochastic Optimization
• Search and Optimization: Non-convex Optimization

Application Category

• Security and Privacy: Data transparency and provenance
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems

Abstract

In this paper, we study the problem of privacy audit in one run and show that our method achieves tight audit results for various differentially private protocols. This includes obtaining tight results for auditing $(varepsilon,δ)$-DP algorithms where all previous work fails to achieve in any parameter setups. We first formulate a framework for privacy audit extit{in one run} with refinement compared with previous work. Then, based on modeling privacy by the $f$-DP formulation, we study the implications of our framework to obtain a theoretically justified lower bound for privacy audit. In the experiment, we compare with previous work and show that our audit method outperforms the rest in auditing various differentially private algorithms. We also provide experiments that give contrasting conclusions to previous work on the parameter settings for privacy audits in one run.

Problem

• Achieving tight privacy audit in one run
• Auditing (ε,δ)-DP algorithms with tight results
• Providing theoretically justified lower bounds for privacy audit

Innovation

• One-run privacy audit framework with refinement
• Theoretically justified lower bound via f-DP
• Outperforms prior work across diverse DP algorithms