一种新的半监督熵模糊聚类算法  

A Novel Semi-supervised Entropy Fuzzy Clustering Algorithm

作　　者：庄美美[1] 张景 ZHUANG Meimei;ZHANG Jing(Quanzhou Branch,Fujian Radio and Television University,Fujian 362000,China;School of Mathematics and Computer Science,Quanzhou Normal University,Fujian 362000,China)

机构地区：[1]福建广播电视大学泉州分校,福建泉州362000 [2]泉州师范学院数学与计算机科学学院,福建泉州362000

出　　处：《泉州师范学院学报》2019年第2期39-45,共7页Journal of Quanzhou Normal University

基　　金：国家大学生创新创业训练项目(20180399037)

摘　　要：结合熵值和半监督模糊聚类两种概念,提出了半监督熵模糊聚类算法.该算法在目标函数中加入了反映划分矩阵的熵值项和已标记数据标签的监督项,使得聚类算法可以同时兼顾划分矩阵的模糊性和已知样本的类别信息.其迭代求解公式具有较为简洁的形式.将聚类方法应用于文本识别及其他真实数据库,实验结果表明,文章提出的半监督聚类算法和传统的聚类算法相比具有更好的聚类效果.Combining the concepts of entropy and semi-supervised fuzzy clustering,a semi-supervised entropy fuzzy clustering algorithm is proposed in this paper.This algorithm adds the entropic term and the supervisory term in its objective function to reflect partition entropy and labeled data information so that the clustering algorithm can take into account both the ambiguity of partition matrix and the class information of labeled samples.Its iterative solution formula has a concise form.This clustering method is applied to text recognition and other real databases.The experimental results show that the semi-supervised clustering algorithm proposed in this paper has better clustering effect than the traditional clustering algorithms.

关 键 词：熵值 模糊聚类 半监督 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]