文章基本信息

标题：LP-NORM SVM BASED ON REWEIGHED MINIMIZATION WITH POSITIVE DAMPED ITEM
本地全文：下载
作者：HUANG HAILONG ; LIU JIANWEI ; LIU ZEYU 等
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2013
卷号：52
期号：3
出版社：Journal of Theoretical and Applied
摘要：A new kind of reweighed Lp-norm SVM, which solves simultaneously two major problems pattern classification and feature selection, based on positive damped item is proposed. The proposed algorithms attempts to select important features among the originally given plausible features, while maintaining the minimum error rate. The resulting value of variable p is not only related to the classification error rate but also connected to the degree of importance of feature. A convergence proof of this reweighed procedure is included and an efficient stopping criterion is employed. Different sets of experiments are conducted on the classification and feature selection tasks, and results compared with L2-norm SVM, and L1-norm SVM show that, our Lp-norm SVM algorithm is superior to these algorithm on both artificial datasets and real-world problems of analyzing DNA microarray data.
关键词：Reweighed Minimization; P Norm; Feature Selection; Prediction Error Rate