Oneshot Differentially Private Top-k Selection

Gang Qiao, Weijie Su, Li Zhang
Proceedings of the 38th International Conference on Machine Learning, PMLR 139:8672-8681, 2021.

Abstract

Being able to efficiently and accurately select the top-$k$ elements with differential privacy is an integral component of various private data analysis tasks. In this paper, we present the oneshot Laplace mechanism, which generalizes the well-known Report Noisy Max \cite{dwork2014algorithmic} mechanism to reporting noisy top-$k$ elements. We show that the oneshot Laplace mechanism with a noise level of $\widetilde{O}(\sqrt{k}/\eps)$ is approximately differentially private. Compared to the previous peeling approach of running Report Noisy Max $k$ times, the oneshot Laplace mechanism only adds noises and computes the top $k$ elements once, hence much more efficient for large $k$. In addition, our proof of privacy relies on a novel coupling technique that bypasses the composition theorems so without the linear dependence on $k$ which is inherent to various composition theorems. Finally, we present a novel application of efficient top-$k$ selection in the classical problem of ranking from pairwise comparisons.

Cite this Paper


BibTeX
@InProceedings{pmlr-v139-qiao21b, title = {Oneshot Differentially Private Top-k Selection}, author = {Qiao, Gang and Su, Weijie and Zhang, Li}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, pages = {8672--8681}, year = {2021}, editor = {Meila, Marina and Zhang, Tong}, volume = {139}, series = {Proceedings of Machine Learning Research}, month = {18--24 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v139/qiao21b/qiao21b.pdf}, url = {https://proceedings.mlr.press/v139/qiao21b.html}, abstract = {Being able to efficiently and accurately select the top-$k$ elements with differential privacy is an integral component of various private data analysis tasks. In this paper, we present the oneshot Laplace mechanism, which generalizes the well-known Report Noisy Max \cite{dwork2014algorithmic} mechanism to reporting noisy top-$k$ elements. We show that the oneshot Laplace mechanism with a noise level of $\widetilde{O}(\sqrt{k}/\eps)$ is approximately differentially private. Compared to the previous peeling approach of running Report Noisy Max $k$ times, the oneshot Laplace mechanism only adds noises and computes the top $k$ elements once, hence much more efficient for large $k$. In addition, our proof of privacy relies on a novel coupling technique that bypasses the composition theorems so without the linear dependence on $k$ which is inherent to various composition theorems. Finally, we present a novel application of efficient top-$k$ selection in the classical problem of ranking from pairwise comparisons.} }
Endnote
%0 Conference Paper %T Oneshot Differentially Private Top-k Selection %A Gang Qiao %A Weijie Su %A Li Zhang %B Proceedings of the 38th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2021 %E Marina Meila %E Tong Zhang %F pmlr-v139-qiao21b %I PMLR %P 8672--8681 %U https://proceedings.mlr.press/v139/qiao21b.html %V 139 %X Being able to efficiently and accurately select the top-$k$ elements with differential privacy is an integral component of various private data analysis tasks. In this paper, we present the oneshot Laplace mechanism, which generalizes the well-known Report Noisy Max \cite{dwork2014algorithmic} mechanism to reporting noisy top-$k$ elements. We show that the oneshot Laplace mechanism with a noise level of $\widetilde{O}(\sqrt{k}/\eps)$ is approximately differentially private. Compared to the previous peeling approach of running Report Noisy Max $k$ times, the oneshot Laplace mechanism only adds noises and computes the top $k$ elements once, hence much more efficient for large $k$. In addition, our proof of privacy relies on a novel coupling technique that bypasses the composition theorems so without the linear dependence on $k$ which is inherent to various composition theorems. Finally, we present a novel application of efficient top-$k$ selection in the classical problem of ranking from pairwise comparisons.
APA
Qiao, G., Su, W. & Zhang, L.. (2021). Oneshot Differentially Private Top-k Selection. Proceedings of the 38th International Conference on Machine Learning, in Proceedings of Machine Learning Research 139:8672-8681 Available from https://proceedings.mlr.press/v139/qiao21b.html.

Related Material