The knockoff filter for FDR control in group-sparse and multitask\n regression

2016/02/10 by Ran Dai, Rina Foygel Barber, Dai, Ran +1 · 3 citations
Computer Science · Engineering · Mathematics · #FOS: Computer and information sciences #Image and Signal Denoising Methods #Methodology (stat.ME) #Sparse and Compressive Sensing Techniques #Statistical Methods and Inference #Survey Sampling and Estimation Techniques

paper · pdf · doi:10.48550/arxiv.1602.03589

openalex publication_date 2016/02/10 · openalex created_date 2019/06/27 · openalex updated_date 2026/07/28

Abstract

We propose the group knockoff filter, a method for false discovery rate\ncontrol in a linear regression setting where the features are grouped, and we\nwould like to select a set of relevant groups which have a nonzero effect on\nthe response. By considering the set of true and false discoveries at the group\nlevel, this method gains power relative to sparse regression methods. We also\napply our method to the multitask regression problem where multiple response\nvariables share similar sparsity patterns across the set of possible features.\nEmpirically, the group knockoff filter successfully controls false discoveries\nat the group level in both settings, with substantially more discoveries made\nby leveraging the group structure.\n

Cited by

• Gold after Randomized Sand: Model-X Split Knockoffs for Controlled Transformation Selection
• Knockoff-Guided Compressive Sensing: A Statistical Machine Learning Framework for Support-Assured Signal Recovery
• Model-free High Dimensional Mediator Selection with False Discovery Rate Control

Related

• Controlling the FDR in variable selection via multiple knockoffs
• One-at-a-time knockoffs: controlled false discovery rate with higher power
• A Powerful and Precise Feature-level Filter using Group Knockoffs
• Communication-Efficient False Discovery Rate Control via Knockoff Aggregation
• The Terminating-Knockoff Filter: Fast High-Dimensional Variable Selection with False Discovery Rate Control