Concurrent Composition for Interactive Differential Privacy with Adaptive Privacy-Loss Parameters

TL;DR

This paper establishes that the privacy guarantees of noninteractive differentially private mechanisms extend to concurrent compositions of interactive mechanisms under various privacy definitions, supporting full adaptivity.

Contribution

It proves that privacy filters and odometers for noninteractive mechanisms also apply to concurrent interactive mechanisms with adaptive privacy-loss parameters, providing a theoretical foundation for full adaptivity.

Findings

Privacy guarantees extend to concurrent interactive mechanisms.

Theoretical foundation for adaptive composition in interactive DP.

Practical implementation provided for deployment.

Abstract

In this paper, we study the concurrent composition of interactive mechanisms with adaptively chosen privacy-loss parameters. In this setting, the adversary can interleave queries to existing interactive mechanisms, as well as create new ones. We prove that every valid privacy filter and odometer for noninteractive mechanisms extends to the concurrent composition of interactive mechanisms if privacy loss is measured using $(\epsilon, \delta)$-DP, $f$-DP, or R\'enyi DP of fixed order. Our results offer strong theoretical foundations for enabling full adaptivity in composing differentially private interactive mechanisms, showing that concurrency does not affect the privacy guarantees. We also provide an implementation for users to deploy in practice.