Improving the Gaussian Mechanism for Differential Privacy: Analytical Calibration and Optimal Denoising

Abstract

The Gaussian mechanism is an essential building block used in multitude of differentially private data analysis algorithms. In this paper we revisit the Gaussian mechanism and show that the original analysis has several important limitations. Our analysis reveals that the variance formula for the original mechanism is far from tight in the high privacy regime ($\epsilon \to 0$) and it cannot be extended to the low privacy regime ($\epsilon \to \infty$). We address these limitations by developing an optimal Gaussian mechanism whose variance is calibrated directly using the Gaussian cumulative density function instead of a tail bound approximation. We also propose to equip the Gaussian mechanism with a post-processing step based on adaptive estimation techniques by leveraging that the distribution of the perturbation is known. Our experiments show that analytical calibration removes at least a third of the variance of the noise compared to the classical Gaussian mechanism, and that denoising dramatically improves the accuracy of the Gaussian mechanism in the high-dimensional regime.

To appear at the 35th International Conference on Machine Learning (ICML), 2018

References in corpus (1)

• Dirichlet draws are sparse with high probability

Cited by in corpus (13)

• How to DP-fy ML: A Practical Guide to Machine Learning with Differential Privacy
• Gaussian Differential Privacy
• The Discrete Gaussian for Differential Privacy
• Privacy Impact on Generalized Nash Equilibrium in Peer-to-Peer Electricity Market
• Privacy and Utility Tradeoff in Approximate Differential Privacy
• Quantifying Membership Inference Vulnerability via Generalization Gap and Other Model Metrics
• On Mitigating the Utility-Loss in Differentially Private Learning: A new Perspective by a Geometrically Inspired Kernel Approach
• Differentially private depth functions and their associated medians
• When Differential Privacy Meets Interpretability: A Case Study
• Towards Distributed Privacy-Preserving Prediction
• Photonic Differential Privacy with Direct Feedback Alignment
• Differentially Private Synthetic Data Release for Topics API Outputs
• Differentially Private Multivariate Statistics with an Application to Contingency Table Analysis