Kernel stein certificates improve privacy checks for specific tasks
Task-aware Kernel Stein certificates for Rényi differential privacy
Cryptography and SecurityMachine Learning
Summary
Protecting privacy when sharing data can be tricky, especially when using advanced privacy methods like Rènyi differential privacy. The authors develop a new way to check these privacy guarantees using mathematical tools called kernel Stein discrepancies. Their approach also customizes privacy checks to focus on what matters for the specific task, making the protection more precise. They show exact results for common privacy methods and illustrate how this works using real data.
What this means in practice
- •For data privacy teams: Provide precise privacy guarantees tailored to specific data tasks beyond standard models.
- •For machine learning engineers: Generate task-specific privacy certificates when deploying privacy-preserving machine learning models using Rènyi differential privacy.
Tested on one dataset.
Authors
Leoni Carla Wirth, Tara Trautwein, Gesine Reinert
Abstract
Rènyi differential privacy, based on Rènyi divergence, has become a notion of choice for privacy guarantees. However, outside the Gaussian setting, Rènyi divergence can be difficult to compute. In particular, Rènyi differential privacy guarantees that exploit task-specific information, as in Rènyi Pufferfish privacy, can be challenging. Here we use kernel Stein discrepancies based on escort distributions to provide certificates for Rènyi differential privacy. By designing the kernel to focus on the relevant privacy directions, we also derive task-aware privacy certificates. We provide exact identities for Gaussian and Laplace mechanisms and bounds for exponential mechanisms, with separate task-aware guarantees. A real data set example illustrates how the protected comparison set and the released subspace jointly determine the resulting certificates.