DMCA
Sparse regression (2013)
BibTeX
@MISC{A13sparseregression,
author = {Kai Zhang A and Qiaojun Wang B and Liang Lan C and Yu Sun D and Ivan Marsic B and Graph Laplacian},
title = {Sparse regression},
year = {2013}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Low-rank approximation a b s t r a c t Advances of modern science and engineering lead to unprecedented amount of data for information processing. Of particular interest is the semi-supervised learning, where very few training samples are available among large volumes of unlabeled data. Graph-based algorithms using Laplacian regularization have achieved state-of-the-art performance, but can induce huge memory and computational costs. In this paper, we introduce L1-norm penalization on the low-rank factorized kernel for efficient, globally optimal model selection in graph-based semi-supervised learning. An important novelty is that our formulation can be transformed to a standard LASSO regression. On one hand, this makes it possible to employ advanced sparse solvers to handle large scale problems; on the other hand, a globally optimal subset of basis can be chosen adaptively given desired strength of penalizing model complexity, in contrast to some current endeavors that pre-determine the basis without coupling it with the learning task. Our algorithm performs competitively with state-of-the-art algorithms on a variety of benchmark data sets. In particular, it is orders of magnitude faster than exact algorithms and achieves a good trade-off between accuracy and scalability. & 2013 Elsevier B.V. All rights reserved.
Keyphrases
optimal model selection computational cost advanced sparse solver l1-norm penalization information processing low-rank approximation training sample graph-based algorithm modern science unlabeled data semi-supervised learning current endeavor exact algorithm unprecedented amount elsevier b.v. right standard lasso regression laplacian regularization state-of-the-art algorithm algorithm performs optimal subset graph-based semi-supervised learning model complexity learning task large scale problem important novelty large volume particular interest state-of-the-art performance engineering lead benchmark data set low-rank factorized kernel good trade-off desired strength huge memory
