Privacy Auditing with One (1) Training Run

2023/05/15 by Thomas Steinke, Milad Nasr, Steinke, Thomas +3 · 17 citations
Computer Science · Mathematics · #Privacy-Preserving Technologies in Data #Statistical Methods and Inference #Stochastic Gradient Optimization Techniques

paper · pdf · doi:10.48550/arxiv.2305.08846

Abstract

We propose a scheme for auditing differentially private machine learning systems with a single training run. This exploits the parallelism of being able to add or remove multiple training examples independently. We analyze this using the connection between differential privacy and statistical generalization, which avoids the cost of group privacy. Our auditing scheme requires minimal assumptions about the algorithm and can be applied in the black-box or white-box setting.

Cited by

• ContextLeak: Auditing Leakage in Private In-Context Learning Methods
• Black-Box Auditing of Quantum Model: Lifted Differential Privacy with Quantum Canaries
• Efficient Public Verification of Private ML via Regularization
• Beyond Membership: Limitations of Add/Remove Adjacency in Differential Privacy
• Observational Auditing of Label Privacy
• Setting ε is not the Issue in Differential Privacy
• PrivacyGuard: A Modular Framework for Privacy Auditing in Machine Learning
• Quantifying Information Disclosure During Gradient Descent Using Gradient Uniqueness
• f-INE: A Hypothesis Testing Framework for Estimating Influence under Training Randomness
• On the Fairness of Privacy Protection: Measuring and Mitigating the Disparity of Group Privacy Risks for Differentially Private Machine Learning
• Train Once, Answer All: Many Pretraining Experiments for the Cost of One
• Adversarial machine learning :
• Tight Privacy Audit in One Run
• Auditing Approximate Machine Unlearning for Differentially Private Models
• Towards Reliable and Generalizable Differentially Private Machine Learning (Extended Version)
• The Surprising Effectiveness of Membership Inference with Simple N-Gram Coverage
• RecPS: Privacy Risk Scoring for Recommender Systems

Related

• How Well Can Differential Privacy Be Audited in One Run?
• Tight Auditing of Differentially Private Machine Learning
• A General Framework for Auditing Differentially Private Machine Learning
• Debugging Differential Privacy: A Case Study for Privacy Auditing
• A General Approach to Adding Differential Privacy to Iterative Training Procedures