摘要
针对传统的k-类支持向量机(SVM)算法对数据进行多分类时存在的特征变量间信息重叠、模型复杂法(度M高CC、)分对类同精类度别低中这的一特系征列变问量题进,文行章赋提权出,用使得用到灰的色综关合联变聚量类(建GR立C k)-对类特SV征M变模量型进,行给分出类了,一并种用改复进相的关k系-数类SVM多分类算法。实证分析表明,该算法的分类效果优于传统算法。
Addressing a series of problems caused by the k-class Support Vector Machine(SVM)algorithm in classifying data into multiple categories,such as information overlapping between feature variables,high model complexity and low classification accuracy,this paper proposes the method of employing Grey Correlation Clustering(GRC)to classify feature variables,using Multiple Correlation Coefficient method(MCC)to empower feature variables in the same category,and adopting the obtained comprehensive variables to establish the k-class SVM model.The paper also presents an improved k-class SVM multi-classification algorithm.The empirical analysis shows that the classification effect of the proposed algorithm is better than that of the traditional algorithm.
作者
谭馨
邓光明
Tan Xin;Deng Guangming(College of Science,Guilin University of Technology,Guilin Guangxi 541006,China;Institute of Applied Statistics,Guilin University of Technology,Guilin Guangxi 541006,China)
出处
《统计与决策》
CSSCI
北大核心
2020年第22期10-14,共5页
Statistics & Decision
基金
国家自然科学基金资助项目(61563013)
广西自然科学基金资助项目(2018GXNSFAA294131)。
关键词
k-类SVM算法
灰色关联聚类
复相关系数法
多分类
k-class support vector machine algorithm
grey relation clustering
multiple correlation coefficient method
multi-classification
