Individualized Privacy Accounting via Subsampling with Applications in Combinatorial Optimization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ghazi, Badih, Kamath, Pritish, Kumar, Ravi, Manurangsi, Pasin, Sealfon, Adam
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914815420661760
author Ghazi, Badih
Kamath, Pritish
Kumar, Ravi
Manurangsi, Pasin
Sealfon, Adam
author_facet Ghazi, Badih
Kamath, Pritish
Kumar, Ravi
Manurangsi, Pasin
Sealfon, Adam
contents In this work, we give a new technique for analyzing individualized privacy accounting via the following simple observation: if an algorithm is one-sided add-DP, then its subsampled variant satisfies two-sided DP. From this, we obtain several improved algorithms for private combinatorial optimization problems, including decomposable submodular maximization and set cover. Our error guarantees are asymptotically tight and our algorithm satisfies pure-DP while previously known algorithms (Gupta et al., 2010; Chaturvedi et al., 2021) are approximate-DP. We also show an application of our technique beyond combinatorial optimization by giving a pure-DP algorithm for the shifting heavy hitter problem in a stream; previously, only an approximateDP algorithm was known (Kaplan et al., 2021; Cohen & Lyu, 2023).
format Preprint
id arxiv_https___arxiv_org_abs_2405_18534
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Individualized Privacy Accounting via Subsampling with Applications in Combinatorial Optimization
Ghazi, Badih
Kamath, Pritish
Kumar, Ravi
Manurangsi, Pasin
Sealfon, Adam
Data Structures and Algorithms
Cryptography and Security
In this work, we give a new technique for analyzing individualized privacy accounting via the following simple observation: if an algorithm is one-sided add-DP, then its subsampled variant satisfies two-sided DP. From this, we obtain several improved algorithms for private combinatorial optimization problems, including decomposable submodular maximization and set cover. Our error guarantees are asymptotically tight and our algorithm satisfies pure-DP while previously known algorithms (Gupta et al., 2010; Chaturvedi et al., 2021) are approximate-DP. We also show an application of our technique beyond combinatorial optimization by giving a pure-DP algorithm for the shifting heavy hitter problem in a stream; previously, only an approximateDP algorithm was known (Kaplan et al., 2021; Cohen & Lyu, 2023).
title Individualized Privacy Accounting via Subsampling with Applications in Combinatorial Optimization
topic Data Structures and Algorithms
Cryptography and Security
url https://arxiv.org/abs/2405.18534