TL;DR
This paper explores the accuracy of locally private histograms across all privacy regimes, especially medium and low privacy levels, providing near-tight bounds and improved protocols with better communication efficiency.
Contribution
It establishes near-tight error bounds for locally private histograms in medium-to-low privacy regimes and introduces a shuffle model protocol with improved message complexity.
Findings
Achieves near-tight bounds on $\,\ell_ extinfty$ error in medium-to-low privacy regimes.
Provides a shuffle model protocol with accuracy comparable to existing methods but with reduced communication.
Empirically compares algorithms across all privacy regimes to evaluate practical performance.
Abstract
Frequency estimation, a.k.a. histograms, is a workhorse of data analysis, and as such has been thoroughly studied under differentially privacy. In particular, computing histograms in the \emph{local} model of privacy has been the focus of a fruitful recent line of work, and various algorithms have been proposed, achieving the order-optimal error in the high-privacy (small ) regime while balancing other considerations such as time- and communication-efficiency. However, to the best of our knowledge, the picture is much less clear when it comes to the medium- or low-privacy regime (large ), despite its increased relevance in practice. In this paper, we investigate locally private histograms, and the very related distribution learning task, in this medium-to-low privacy regime, and establish near-tight (and somewhat unexpected) bounds on the…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
Locally Private Histograms in All Privacy Regimes· youtube
