Improving knockoffs with conditional calibration
Yixiang Luo, William Fithian, Lihua Lei
The Annals of Statistics · 2025 · Volume 53 · Issue 5
Abstract
The knockoff filter of (Ann. Statist. 43 (2015) 2055–2085) is a flexible framework for multiple testing in supervised learning models, based on introducing synthetic predictor variables to control the false discovery rate (FDR). Using the conditional calibration framework of (Ann. Statist. 50 (2022) 3091–3118), we introduce the calibrated knockoff procedure, a method that uniformly improves the power of any fixed-X or model-X knockoff procedure. We show theoretically and empirically that the improvement is especially notable in two contexts where knockoff methods can be nearly powerless: when the rejection set is small, and when the structure of the design matrix in fixed-X knockoffs prevents us from constructing good knockoff variables. In these contexts, calibrated knockoffs even outperform competing FDR-controlling methods like the (dependence-adjusted) Benjamini–Hochberg procedure in many scenarios.
Keywords
62H15, 62J15, Knockoffs, linear model, Multiple hypotheses testing
Enable JavaScript for citation tools and additional paper interactions.