Numerical Composition of Differential PrivacyDownload PDF

Published: 09 Nov 2021, Last Modified: 22 Oct 2023NeurIPS 2021 SpotlightReaders: Everyone
Keywords: Differential Privacy, Composition
TL;DR: A fast and accurate algorithm to compose privacy guarantees of differentially private algorithms.
Abstract: We give a fast algorithm to compose privacy guarantees of differentially private (DP) algorithms to arbitrary accuracy. Our method is based on the notion of privacy loss random variables to quantify the privacy loss of DP algorithms. The running time and memory needed for our algorithm to approximate the privacy curve of a DP algorithm composed with itself $k$ times is $\tilde{O}(\sqrt{k})$. This improves over the best prior method by Koskela et al. (2020) which requires $\tilde{\Omega}(k^{1.5})$ running time. We demonstrate the utility of our algorithm by accurately computing the privacy loss of DP-SGD algorithm of Abadi et al. (2016) and showing that our algorithm speeds up the privacy computations by a few orders of magnitude compared to prior work, while maintaining similar accuracy.
Code Of Conduct: I certify that all co-authors of this work have read and commit to adhering to the NeurIPS Statement on Ethics, Fairness, Inclusivity, and Code of Conduct.
Supplementary Material: pdf
Code: https://github.com/microsoft/prv_accountant
Community Implementations: [![CatalyzeX](/images/catalyzex_icon.svg) 1 code implementation](https://www.catalyzex.com/paper/arxiv:2106.02848/code)
14 Replies

Loading