PANORAMIA: Privacy Auditing of Machine Learning Models without Retraining

2024/02/12 by Mishaal Kazmi, Kazmi, Mishaal, Hadrien Lautraite +13 · 3 citations
Computer Science · Business, Management and Accounting · #Privacy-Preserving Technologies in Data #Explainable Artificial Intelligence (XAI) #Big Data and Business Intelligence

paper · pdf · doi:10.48550/arxiv.2402.09477

Abstract

We present PANORAMIA, a privacy leakage measurement framework for machine learning models that relies on membership inference attacks using generated data as non-members. By relying on generated non-member data, PANORAMIA eliminates the common dependency of privacy measurement tools on in-distribution non-member data. As a result, PANORAMIA does not modify the model, training data, or training process, and only requires access to a subset of the training data. We evaluate PANORAMIA on ML models for image and tabular data classification, as well as on large-scale language models.

Cited by

• ContextLeak: Auditing Leakage in Private In-Context Learning Methods
• Observational Auditing of Label Privacy
• Uncovering Pretraining Code in LLMs: A Syntax-Aware Attribution Approach

Related

• Membership Privacy for Machine Learning Models Through Knowledge Transfer
• Range Membership Inference Attacks
• SHAPr: An Efficient and Versatile Membership Privacy Risk Metric for Machine Learning
• ML Privacy Meter: Aiding Regulatory Compliance by Quantifying the Privacy Risks of Machine Learning
• Membership Inference Attacks Beyond Overfitting