The Gaussian mechanism is an essential building block used in multitude of differentially private data analysis algorithms. In this talk I will revisit the classical analysis of the Gaussian mechanism and show it has several important limitations. For example, our analysis reveals that the variance formula for the original mechanism is far from tight in the high privacy regime and that it cannot be extended to the low privacy regime. We address these limitations by developing a new Gaussian mechanism whose variance is optimally calibrated by solving an equation involving the Gaussian cumulative density function. Our analysis side-steps the use of tail bounds approximations and relies on a novel characterisation of differential privacy that might be of independent interest. We numerically show that analytical calibration removes at least a third of the variance of the noise compared to the classical Gaussian mechanism. We also propose to equip the Gaussian mechanism with a post-processing step based on adaptive denoising estimators by leveraging that the variance of the perturbation is known. Experiments with synthetic and real data show that this denoising step yields dramatic accuracy improvements in the high-dimensional regime.
Based on joint work with Y.-X. Wang to appear at ICML 2018.
Biography: Borja Balle is currently a Machine Learning Scientist at Amazon Research in Cambridge. Before joining Amazon, Borja was a lecturer at Lancaster University (2015-2017), a postdoctoral fellow at McGill University (2013-2015), and a graduate student at Universitat Politecnica de Catalunya where he obtained his PhD in 2013. His main research interest is in privacy-preserving machine learning, including the use of differential privacy and multi-party computation in distributed learning problems, and the foundations of privacy-aware data science. More info: https://borjaballe.github.io