Estimation of Graphical Models using the ℓ₁,₂ Norm
Oct 1, 2018·,·
1 min read
Khai Chiong
Roger Moon
Abstract
We propose a method for estimating graphical models using the mixed ℓ₁,₂ norm penalty to induce group sparsity in precision matrices. The approach facilitates consistent estimation of network structures in high-dimensional settings.
Type
Publication
Econometrics Journal, Volume 21, Issue 3, pages 247–263, October 2018
In this article, Chiong and Moon develop estimation techniques for graphical models based on the ℓ₁,₂ norm penalty. The method promotes group sparsity in the inverse covariance matrix, allowing practitioners to uncover network structures in data sets with many variables. The technique has applications in econometrics and machine learning.