一种基于模拟退火和遗传算法的模糊聚类方法  

A Fuzzy Clustering Algorithm Based on Simulated Annealing and Genetic Algorithm

作　　者：白莉媛[1] 胡声艳[1] 刘素华[1] 

机构地区：[1]河南工业大学计算机科学系,郑州450052

出　　处：《计算机工程与应用》2005年第9期56-58,共3页Computer Engineering and Applications

基　　金：河南省科技攻关项目资助(编号:0224010011)

摘　　要：首先,对模糊C-均值聚类算法做了简要分析和评论,根据其特点,提出了一种基于模拟退火和遗传算法的聚类分析方法,算法中采用了适合于模糊聚类的树型编码方案。实验表明,该算法可克服系统对数据集及初始聚类中心的敏感性,避免陷入局部极小,在模式识别、数据挖掘等领域有着广泛的应用前景。This paper firstly describes the fuzzy C-means clustering algorithm.Then,according to its characteristic,an improved clustering analysis method based on simulated annealing and genetic algorithm is proposed.A tree encoding scheme which is proper for the fuzzy clustering algorithm is determined.Experiment results show that the improved algorithm give us more satisfactory results by escaping from local minima regardless of the data set and the selection of the initial clustering centers.It can be used in many fields such as pattern recognition and data mining.

关 键 词：模糊聚类 遗传算法 模拟退火算法 树型编码 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]