摘要
提出一种基于属性和距离加权的K近邻特征选择方法.该方法在计算样本类别时既考虑每个特征的重要程度,又考虑近邻样本的距离,使用遗传算法搜索最优特征权重向量.将该方法与已有的3种特征选择方法MIFS、DISR和CIFE在6个公开的数据集上进行比较,实验结果表明该方法是有效的,且可以提高分类性能.
A feature selection method based on attribute and distance weighted K-nearest neighbors is proposed. The method not only takes into account the different distance of nearest neighbor samples,but also takes into account the different degree of importance of features when calculating the categories of samples,and the optimal feature weight vector is searched by using genetic algorithm. The method is compared with the existing 3 feature selection methods MIFS,DISR and CIFE on 6 public datasets,and the experimental results show that the proposed method is effective and can improve the classification perfor-mance.
作者
李双杰
张开翔
王士栋
王淑琴
LI Shuangjie;ZHANG Kaixiang;WANG Shidong;WANG Shuqin(College of Computer and Information Engineering,Tianjin Normal University,Tianjin 300387,China)
出处
《天津师范大学学报（自然科学版）》
CAS
北大核心
2020年第2期63-67,共5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(15JCYBJC4600,19JCZDJC35100)
天津市科技计划资助项目(16ZLZDZF00150).
关键词
特征选择
K近邻
遗传算法
实数编码
feature selection
K-nearest neighbors
genetic algorithms
real coding
