基于互信息与判别结构互补的特征选择  

Feature selection based on mutual information and complementary discriminant structure

作　　者：张开翔 李双杰 王小刚 李雅丽 陈倩茹 王淑琴[1] ZHANG Kaixiang;LI Shuangjie;WANG Xiaogang;LI Yali;CHEN Qianru;WANG Shuqin(College of Computer and Information Engineering,Tianjin Normal University,Tianjin 300387,China)

机构地区：[1]天津师范大学计算机与信息工程学院,天津300387

出　　处：《天津师范大学学报（自然科学版）》2021年第1期63-68,共6页Journal of Tianjin Normal University：Natural Science Edition

基　　金：国家自然科学基金资助项目(61070089);天津市应用基础与前沿技术研究计划重点资助项目(15JCYBJC4600);天津市科技计划资助项目(19JCZDJC35100).

摘　　要：提出一种基于子问题分类能力的特征选择算法,利用互信息度量特征对子问题的分类能力,根据特征的判别结构互补确定目标特征子集.该算法既会选择总分类能力强的特征,也会选择子问题分类能力强的特征.将该算法与已有的6个特征选择算法在6个公开的数据集上进行比较实验,结果表明,所提算法的性能优于其他特征选择算法.A feature selection algorithm based on the classification ability of subproblems is proposed,which uses mutual infor-mation to measure the classification ability of feature for subproblems,and determines the target feature subset according to the complementary discriminant structure.In the algorithm,not only the features with strong ability for total classification,but also the features with strong ability for subproblem classification can be selected.The algorithm is compared with six existing feature selec-tion algorithms on six public datasets.And the experimental results show that the performance of the proposed algorithm is superior to other feature selection algorithms.

关 键 词：特征选择 互信息 判别结构互补 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]