Least Absolute Gradient Selector: Statistical Regression via Pseudo-Hard Thresholding

TL;DR

The paper introduces LAGS, a convex optimization method that mimics hard-thresholding for variable selection in linear models, achieving model sparsity and consistency with efficient computation.

Contribution

LAGS is a novel convex program that exhibits pseudo-hard thresholding, combining the benefits of $l_0$ regularization with computational efficiency.

Findings

LAGS demonstrates discrete selection behavior similar to $l_0$ regularization.

LAGS is consistent and can identify the true model asymptotically.

Numerical simulations show LAGS outperforms soft-thresholding in MSE and model parsimony.

Abstract

Variable selection in linear models plays a pivotal role in modern statistics. Hard-thresholding methods such as $l_0$ regularization are theoretically ideal but computationally infeasible. In this paper, we propose a new approach, called the LAGS, short for "least absulute gradient selector", to this challenging yet interesting problem by mimicking the discrete selection process of $l_0$ regularization. To estimate $\beta$ under the influence of noise, we consider, nevertheless, the following convex program [\hat{\beta} = \textrm{arg min}\frac{1}{n}\|X^{T}(y - X\beta)\|_1 + \lambda_n\sum_{i = 1}^pw_i(y;X;n)|\beta_i|] $\lambda_n > 0$ controls the sparsity and $w_i > 0$ dependent on $y, X$ and $n$ is the weights on different $\beta_i$; $n$ is the sample size. Surprisingly, we shall show in the paper, both geometrically and analytically, that LAGS enjoys two attractive properties: (1) LAGS demonstrates discrete selection behavior and hard thresholding property as $l_0$ regularization by strategically chosen $w_i$, we call this property "pseudo-hard thresholding"; (2) Asymptotically, LAGS is consistent and capable of discovering the true model; nonasymptotically, LAGS is capable of identifying the sparsity in the model and the prediction error of the coefficients is bounded at the noise level up to a logarithmic factor---$\log p$, where $p$ is the number of predictors. Computationally, LAGS can be solved efficiently by convex program routines for its convexity or by simplex algorithm after recasting it into a linear program. The numeric simulation shows that LAGS is superior compared to soft-thresholding methods in terms of mean squared error and parsimony of the model.