计算机应用与软件2017，Vol.34Issue(7)：298-302,324,6.DOI:10.3969/j.issn.1000-386x.2017.07.055
扩展ReliefF的两种多标签特征选择算法
TWO FEATURE SELECTION ALGORITHMS FOR MULTI-LABEL CLASSIFICATION BY EXTENDING RELIEFF
马晶莹 1宣恒农1
作者信息
- 1. 南京财经大学信息工程学院 江苏 南京 210000
- 折叠
摘要
Abstract
The ReliefF feature selection algorithm is limited to single-label data.Concerning this problem, two multi-label feature selection algorithms termed Mult-ReliefF and M-A are proposed.The Mult-ReliefF algorithm redefined the relationship of the nearest neighbors and updating formula of feature weights and added the label contribution to update the feature weight formula.Based on the Mult-ReliefF algorithm, combining with neighborhood rough set to achieve better effect of the feature dimension reduction, secondary reduction algorithm M-A is proposed by also adopting the ML-KNN sorting algorithm.Experimental results on data sets show that Mult-ReliefF algorithm can improve classification effect and M-A can obtain smaller feature subset.关键词
多标签分类/特征选择/数据降维/ReliefF/邻域粗糙集Key words
Multi-label classification/ Feature selection/ Attribute reduction/ ReliefF/ Neighborhood rough sets分类
信息技术与安全科学引用本文复制引用
马晶莹,宣恒农..扩展ReliefF的两种多标签特征选择算法[J].计算机应用与软件,2017,34(7):298-302,324,6.
