Speaker: Xintao Xia, Graduate Student, Iowa State University

Title: Adaptive False Discovery Rate Control with Privacy Guarantee

Abstract: Differentially private multiple-testing procedures can protect the information of individual hypothesis tests while guaranteeing a small fraction of false discoveries. In this paper, we propose a differentially private adaptive FDR control method that can control the classic FDR metric exactly at a user-specified level α with privacy guarantee, which is a non-trivial improvement compared to the DP-BH method proposed in (Dwork, et. 2021). Our analysis is based on two key insights: 1) a novel p-value transformation that preserves both privacy and the mirror conservative property, and 2) a mirror peeling algorithm that allows the construction of the filtration and application of the optimal stopping technique. Numerical studies demonstrate that the proposed DP-AdaPT performs better compared to the existing differentially private FDR control methods. Compared to the original AdaPT, it only incurs a small accuracy loss but also significantly reduces the computation cost.