A novel L1/2 regularization shooting method for Cox's proportional hazards model

被引：2

作者：

Luan, Xin-Ze ^{[1
]}

Liang, Yong ^{[1
]}

Liu, Cheng ^{[1
]}

Leung, Kwong-Sak ^{[2
]}

Chan, Tak-Ming ^{[2
]}

Xu, Zong-Ben ^{[3
]}

Zhang, Hai ^{[3
]}

机构：

[1] Macau Univ Sci & Technol, Macau, Peoples R China

[2] Chinese Univ Hong Kong, Hong Kong, Hong Kong, Peoples R China

[3] Xi An Jiao Tong Univ, Xian 710049, Peoples R China

来源：

SOFT COMPUTING | 2014年 / 18卷 / 01期

基金：

中国国家自然科学基金;

关键词：

Variable selection; Cox model; Lasso; L-1/2 regularization shooting algorithm; NONCONCAVE PENALIZED LIKELIHOOD; BAYESIAN VARIABLE SELECTION; ADAPTIVE LASSO; SURVIVAL; REGRESSION;

D O I：

10.1007/s00500-013-1042-6

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Nowadays, a series of methods are based on a L (1) penalty to solve the variable selection problem for a Cox's proportional hazards model. In 2010, Xu et al. have proposed a L (1/2) regularization and proved that the L (1/2) penalty is sparser than the L (1) penalty in linear regression models. In this paper, we propose a novel shooting method for the L (1/2) regularization and apply it on the Cox model for variable selection. The experimental results based on comprehensive simulation studies, real Primary Biliary Cirrhosis and diffuse large B cell lymphoma datasets show that the L (1/2) regularization shooting method performs competitively.

引用

页码：143 / 152

页数：10

共 39 条

[11] PARTIAL LIKELIHOOD [J].