Covariate-Correlated Lasso for Feature Selection
Abstract
Lasso-type variable selection has been increasingly adopted in many applications. In this paper, we propose a covariate-correlated Lasso that selects the covariates correlated more strongly with the response variable. We propose an efficient algorithm to solve this Lasso-type optimization and prove its convergence. Experiments on DNA gene expression data sets show that the selected covariates correlate more strongly with the response variable, and the residual values are decreased, indicating better covariate selection. The selected covariates lead to better classification performance.
Cite
Text
Jiang et al. "Covariate-Correlated Lasso for Feature Selection." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2014. doi:10.1007/978-3-662-44848-9_38Markdown
[Jiang et al. "Covariate-Correlated Lasso for Feature Selection." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2014.](https://mlanthology.org/ecmlpkdd/2014/jiang2014ecmlpkdd-covariatecorrelated/) doi:10.1007/978-3-662-44848-9_38BibTeX
@inproceedings{jiang2014ecmlpkdd-covariatecorrelated,
title = {{Covariate-Correlated Lasso for Feature Selection}},
author = {Jiang, Bo and Ding, Chris H. Q. and Luo, Bin},
booktitle = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
year = {2014},
pages = {595-606},
doi = {10.1007/978-3-662-44848-9_38},
url = {https://mlanthology.org/ecmlpkdd/2014/jiang2014ecmlpkdd-covariatecorrelated/}
}