Recovering The Graphical Structures Via Knockoffs

Keywords

false discovery rate (FDR); Gaussian graphical models; knockoffs; neighborhood selection; Variable selection

Abstract

Learning the dependence structures in Gaussian graphical models is of fundamental importance in many contemporary applications. Despite the fast growing literature, procedures with guaranteed FDR control for recovering the graphical structures are rare. In this paper, we propose a new procedure based on constructing knockoff variables such that the FDR for graph recovery can be controlled nodewisely. The suggested method combines the strengths of FDR control via knockoffs in linear regression settings and neighborhood selection which converts the problem of identifying Gaussian graphical structures into nodewise variable selection. Numerical studies show that the proposed procedure enjoys better statistical power compared with existing methods.

Publication Date

1-1-2018

Publication Title

Procedia Computer Science

Volume

129

Number of Pages

201-207

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

DOI Link

https://doi.org/10.1016/j.procs.2018.03.039

Socpus ID

85047071881 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/85047071881

This document is currently not available here.

Share

COinS