Differentially private partition selection

TL;DR

This paper introduces an optimal differentially private mechanism for partition selection in data analysis, maximizing the number of released partitions while ensuring privacy, especially when each user is associated with a single partition.

Contribution

It proposes a simple, optimal mechanism for differentially private partition selection in the single-partition-per-user setting, advancing the formal study of differentially private set union.

Findings

Maximizes the number of released partitions under differential privacy

Provides implementation considerations for practical deployment

Extends potential applicability to multiple partitions per user

Abstract

Many data analysis operations can be expressed as a GROUP BY query on an unbounded set of partitions, followed by a per-partition aggregation. To make such a query differentially private, adding noise to each aggregation is not enough: we also need to make sure that the set of partitions released is also differentially private. This problem is not new, and it was recently formally introduced as differentially private set union. In this work, we continue this area of study, and focus on the common setting where each user is associated with a single partition. In this setting, we propose a simple, optimal differentially private mechanism that maximizes the number of released partitions. We discuss implementation considerations, as well as the possible extension of this approach to the setting where each user contributes to a fixed, small number of partitions.