Differentially Private Model-X Knockoffs via Johnson-Lindenstrauss Transform

TL;DR

This paper presents a new method for high-dimensional variable selection that guarantees differential privacy and FDR control by privatizing the data using Johnson-Lindenstrauss Transform, preserving statistical power.

Contribution

It introduces a novel privatization approach using Johnson-Lindenstrauss Transform for Model-X knockoffs, enabling privacy-preserving variable selection with theoretical FDR and power guarantees.

Findings

Johnson-Lindenstrauss Transform preserves covariate relationships under privacy constraints.

The proposed method maintains statistical power better than classical noise addition.

Theoretical analysis characterizes privacy-power trade-offs in high-dimensional settings.

Abstract

We introduce a novel privatization framework for high-dimensional controlled variable selection. Our framework enables rigorous False Discovery Rate (FDR) control under differential privacy constraints. While the Model-X knockoff procedure provides FDR guarantees by constructing provably exchangeable ``negative control" features, existing privacy mechanisms like Laplace or Gaussian noise injection disrupt its core exchangeability conditions. Our key innovation lies in privatizing the data knockoff matrix through the Gaussian Johnson-Lindenstrauss Transformation (JLT), a dimension reduction technique that simultaneously preserves covariate relationships through approximate isometry for $(\epsilon,\delta)$-differential privacy. We theoretically characterize both FDR and the power of the proposed private variable selection procedure, in an asymptotic regime. Our theoretical analysis characterizes the role of different factors, such as the JLT's dimension reduction ratio, signal-to-noise ratio, differential privacy parameters, sample size and feature dimension, in shaping the privacy-power trade-off. Our analysis is based on a novel `debiasing technique' for high-dimensional private knockoff procedure. We further establish sufficient conditions under which the power of the proposed procedure converges to one. This work bridges two critical paradigms -- knockoff-based FDR control and private data release -- enabling reliable variable selection in sensitive domains. Our analysis demonstrates that structural privacy preservation through random projections outperforms the classical noise addition mechanism, maintaining statistical power even under strict privacy budgets.