Sharp Composition Bounds for Gaussian Differential Privacy via Edgeworth Expansion

by   Qinqing Zheng, et al.
University of Pennsylvania

Datasets containing sensitive information are often sequentially analyzed by many algorithms and, accordingly, a fundamental question in differential privacy is concerned with how the overall privacy bound degrades under composition. To address this question, we introduce a family of analytical and sharp privacy bounds under composition using the Edgeworth expansion in the framework of the recently proposed f-differential privacy. In contrast to the existing composition theorems using the central limit theorem, our new privacy bounds under composition gain improved tightness by leveraging the refined approximation accuracy of the Edgeworth expansion. Our approach is easy to implement and computationally efficient for any number of compositions. The superiority of these new bounds is confirmed by an asymptotic error analysis and an application to quantifying the overall privacy guarantees of noisy stochastic gradient descent used in training private deep neural networks.


page 1

page 2

page 3

page 4


Analytical Composition of Differential Privacy via the Edgeworth Accountant

Many modern machine learning algorithms are composed of simple private a...

Gaussian Differential Privacy

Differential privacy has seen remarkable success as a rigorous and pract...

Deep Learning with Gaussian Differential Privacy

Deep learning models are often trained on datasets that contain sensitiv...

Identification, Amplification and Measurement: A bridge to Gaussian Differential Privacy

Gaussian differential privacy (GDP) is a single-parameter family of priv...

Fully Adaptive Composition in Differential Privacy

Composition is a key feature of differential privacy. Well-known advance...

Composition in Differential Privacy for General Granularity Notions (Long Version)

The composition theorems of differential privacy (DP) allow data curator...

Privacy Amplification of Iterative Algorithms via Contraction Coefficients

We investigate the framework of privacy amplification by iteration, rece...

Code Repositories

Please sign up or login with your details

Forgot password? Click here to reset