On constrained and regularized high-dimensional regression

By Xiaotong Shen, Wei Pan, Yunzhang Zhu, Hui Zhou. 2013

86 citations recorded by the source corpus.

High-dimensional feature selection has become increasingly crucial for seeking parsimonious models in estimation. For selection consistency, we derive one necessary and sufficient condition formulated on the notion of degree of separation. The minimal degree of separation is necessary for any method to be selection consistent. At a level slightly higher than the minimal degree of separation, selection consistency is achieved by a constrained \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_0$$\end{document}-method and its computational surrogate—the constrained truncated \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_1$$\end{document}-method. This permits up to exponentially many features in the sample size. In other words, these methods are optimal in feature selection against any selection method. In contrast, their regularization counterparts—the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_0$$\end{document}-regularization and truncated \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_1$$\end{document}-regularization methods enable so under slightly stronger assumptions. More importantly, sharper parameter estimation/prediction is realized through such selection, leading to minimax parameter estimation. This, otherwise, is impossible in the absence of a good selection method for high-dimensional analysis.

Subfields: Control and Systems Engineering, Mathematical Physics, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Control and Systems Engineering

  1. Dimitris J. Bertsimas
  2. Angela King
  3. Rahul Mazumder

In Mathematical Physics

  1. Rahul Mazumder

In Statistics and Probability

  1. Rahul Mazumder
  2. Dimitris J. Bertsimas
  3. Angela King
  4. Jack Dunn
  5. Velibor V. Misic
  6. Peter Radchenko
  7. Robert Tibshirani
  8. Trevor J. Hastie

Citations in the corpus