English

Query-Efficient Locally Private Hypothesis Selection via the Scheffe Graph

Data Structures and Algorithms 2026-01-16 v2 Machine Learning Machine Learning

Abstract

We propose an algorithm with improved query-complexity for the problem of hypothesis selection under local differential privacy constraints. Given a set of kk probability distributions QQ, we describe an algorithm that satisfies local differential privacy, performs O~(k3/2)\tilde{O}(k^{3/2}) non-adaptive queries to individuals who each have samples from a probability distribution pp, and outputs a probability distribution from the set QQ which is nearly the closest to pp. Previous algorithms required either Ω(k2)\Omega(k^2) queries or many rounds of interactive queries. Technically, we introduce a new object we dub the Scheff\'e graph, which captures structure of the differences between distributions in QQ, and may be of more broad interest for hypothesis selection tasks.

Keywords

Cite

@article{arxiv.2509.16180,
  title  = {Query-Efficient Locally Private Hypothesis Selection via the Scheffe Graph},
  author = {Gautam Kamath and Alireza F. Pour and Matthew Regehr and David P. Woodruff},
  journal= {arXiv preprint arXiv:2509.16180},
  year   = {2026}
}
R2 v1 2026-07-01T05:46:11.867Z