MIST: L0 Sparse Linear Regression with Momentum
Machine Learning
2015-03-20 v2
Abstract
Significant attention has been given to minimizing a penalized least squares criterion for estimating sparse solutions to large linear systems of equations. The penalty is responsible for inducing sparsity and the natural choice is the so-called norm. In this paper we develop a Momentumized Iterative Shrinkage Thresholding (MIST) algorithm for minimizing the resulting non-convex criterion and prove its convergence to a local minimizer. Simulations on large data sets show superior performance of the proposed method to other methods.
Cite
@article{arxiv.1409.7193,
title = {MIST: L0 Sparse Linear Regression with Momentum},
author = {Goran Marjanovic and Magnus O. Ulfarsson and Alfred O. Hero},
journal= {arXiv preprint arXiv:1409.7193},
year = {2015}
}