期刊
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
卷 289, 期 1, 页码 214-231出版社
ELSEVIER
DOI: 10.1016/j.ejor.2020.06.045
关键词
Analytics; Feature selection; Clustering; Square root fused LASSO; Alternating direction method of multipliers
资金
- National Natural Science Foundation of China [71901109, 61973145, 71861012, 71761016]
- Natural Science Foundation of Jiangxi, China [20181BAB211020]
- Postdoctoral Foundation of Jiangxi Province [2018KY08]
- Scientific Research Fund of Jiangxi Provincial Education Department [GJJ180287, GJJ180247, GJJ190264]
- Human and Social Science Foundation of Jiangxi Province [TJ19202]
The proposed square root fused LASSO method addresses the challenge of tuning regularization parameters in the LASSO method dependent on noise level sigma by combining square root loss function and penalty functions. Theoretical derivations and experiments demonstrate the superiority of this method.
The fused least absolute shrinkage and selection operator (LASSO) simultaneously pursuing the joint sparsity of coefficients and their successive differences has attracted significant attention for analytics purposes. Although it is extensively used, especially when the number of features exceeds the sample size, tuning the regularization parameters, which depends on noise level sigma, is a challenging task since sigma is difficult to estimate accurately. To tackle this problem, in this paper, we propose and study square root fused LASSO, which combines the square root loss function and joint penalty functions. In theory, we show that the proposed method can achieve the same error rate as that of fused LASSO by proving its estimation and prediction error bounds. In addition, the error rate of square root fused LASSO is lower than those of LASSO and square root LASSO via simultaneous feature selection and clustering. The choices of the regularization parameters are also shown to be free of sigma. In terms of computation, this work develops a novel algorithm based on the alternating direction method of multipliers algorithm with theoretical guarantee of its convergence. Experiments on simulation and real-world datasets demonstrate the superiority of square root fused LASSO over fused LASSO and other state-of-the-art feature selection methods. (C) 2020 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据