Skip to main navigation Skip to search Skip to main content

Differentially Private High-Dimensional Approximate Range Counting, Revisited.

Research output: Conference Article in Proceeding or Book/Report chapterArticle in proceedingsResearchpeer-review

Abstract

Locality Sensitive Filters are known for offering a quasi-linear space data structure with rigorous guarantees for the Approximate Near Neighbor search (ANN) problem. Building on Locality Sensitive Filters, we derive a simple data structure for the Approximate Near Neighbor Counting (ANNC) problem under differential privacy (DP). Moreover, we provide a simple analysis leveraging a connection with concomitant statistics and extreme value theory. Our approach produces a simple data structure with a tunable parameter that regulates a trade-off between space-time and utility. Through this trade-off, our data structure achieves the same performance as the recent findings of Andoni et al. (NeurIPS 2023) while offering better utility at the cost of higher space and query time. In addition, we provide a more efficient algorithm under pure ε-DP and elucidate the connection between ANN and differentially private ANNC. As a side result, the paper provides a more compact description and analysis of Locality Sensitive Filters for Fair Near Neighbor Search, improving a previous result in Aumüller et al. (TODS 2022).
Original languageEnglish
Title of host publication6th Symposium on Foundations of Responsible Computing (FORC 2025)
Number of pages24
Volume329
PublisherSchloss Dagstuhl - Leibniz-Zentrum fuer Informatik GmbH
Publication date2025
Pages1-24
ISBN (Print)978-3-95977-367-6
DOIs
Publication statusPublished - 2025
EventFoundations of Responsible Computing - Tresidder Oak Lounge, Stanford, United States
Duration: 4 Jun 20256 Jun 2025
Conference number: 6
https://responsiblecomputing.org/forc-2025/

Conference

ConferenceFoundations of Responsible Computing
Number6
LocationTresidder Oak Lounge
Country/TerritoryUnited States
CityStanford
Period04/06/202506/06/2025
Internet address
SeriesLeibniz International Proceedings in Informatics (LIPIcs)
ISSN1868-8969

Keywords

  • Differential Privacy
  • Locality Sensitive Filters
  • Approximate Range Counting
  • Concominant Statistics

Fingerprint

Dive into the research topics of 'Differentially Private High-Dimensional Approximate Range Counting, Revisited.'. Together they form a unique fingerprint.

Cite this