一种新的聚类有效性函数:模糊划分的模糊熵  

A new clustering effectiveness function: fuzzy entropy of fuzzy partition

作　　者：卿铭[1] 孙晓梅[2] 

机构地区：[1]西南交通大学数学学院,四川成都600031 [2]河南工程学院数学与物理科学系,河南郑州451191

出　　处：《智能系统学报》2015年第1期75-80,共6页CAAI Transactions on Intelligent Systems

基　　金：中央高校基础研究基金资助项目(2682014ZT28)

摘　　要：模糊聚类分析结果是否合理的问题属于模糊聚类有效性判定课题,其核心是模糊聚类有效性函数的构造。文中基于序关系定义了模糊划分模糊熵来描述模糊划分的模糊程度。考虑到现有的一类有效的模糊聚类有效性函数就是基于数据集的模糊划分的,因此文中也用模糊划分的模糊熵作为聚类有效性函数。实验表明,模糊划分的模糊熵作为模糊聚类的有效性函数是合理的、可行的。In this paper,the determination that whether a fuzzy clustering analysis result is reasonable or not is decided by the effectiveness of fuzzy clustering and its core is the construction of fuzzy clustering effectiveness function. This paper proposed a new concept of fuzzy entropy for a fuzzy partition to describe fuzzy degree of a fuzzy partition based on an order relation. Fuzzy entropy for a fuzzy partition is also considered as a clustering effectiveness function because some existing fuzzy clustering effectiveness functions are based on fuzzy partition of data sets. The experiments demonstrated that it is reasonable and practicable to utilize fuzzy entropy for a fuzzy partition as the effectiveness function of a fuzzy clustering.

关 键 词：模糊C均值聚类 模糊划分的模糊熵 聚类有效性 聚类分析 模糊划分 模糊熵 熵函数 模糊集 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]