摘要
文章提出了2种基于佳点集遗传算法的模糊聚类新方法GgaFca和HGgaFca。GgaFca可用于发现指定簇数(c)的聚类中心,具有对初始输入不敏感、收敛快、精度高并可避免早熟的特点;而混合方法HGgaFcm是利用传统模糊c-均值(Fcm)聚类算法对GgaFca聚类结果的进一步提炼,实验结果表明它具有更好的聚类效果和综合性能,可适用于不同数据库下的模糊聚类挖掘研究。
Two new fuzzy clustering approaches, GgaFca and HGgaFca,are proposed based on the good point-set genetic algorithm. It is shown how GgaFca can be used to find the centroid of a user specified number (c) of clusters, which is characterized by inferior sensitivity to initial input, quick convergence, higher accuracy and removal of premature. The hybrid algorithm,HGgaFca,basically uses fuzzy c-means(Fcm) clustering algorithm to refine the clusters formed by GgaFca. The experiment results show by comparison that the hybrid algorithm,HGgaFca,has better general performance and can be applied to various databases for clustering data in data mining.
出处
《合肥工业大学学报（自然科学版）》
CAS
CSCD
北大核心
2005年第4期402-406,共5页
Journal of Hefei University of Technology：Natural Science
基金
安徽省高校自然科学研究资助项目(2005kj095)
关键词
模糊聚类
佳点集
遗传算法
模糊C-均值
fuzzy clustering
good point-set
genetic algorithm
fuzzy c-means(FCM)
