基于遗传算法的K均值聚类分析  

K-Means Clustering Analysis Based on Genetic Algorithm

作　　者：赖玉霞[1] 刘建平[1] 杨国兴[2] 

机构地区：[1]浙江理工大学信息电子学院,杭州310018 [2]浙江天健会计师事务所,杭州310012

出　　处：《计算机工程》2008年第20期200-202,共3页Computer Engineering

摘　　要：传统K均值算法对初始聚类中心敏感,聚类结果随不同的初始输入而波动,容易陷入局部最优值。针对上述问题,该文提出一种基于遗传算法的K均值聚类算法,将K均值算法的局部寻优能力与遗传算法的全局寻优能力相结合,在自适应交叉概率和变异概率的遗传算法中引入K均值操作,以克服传统K均值算法的局部性和对初始中心的敏感性,实验证明,该算法有较好的全局收敛性,聚类效果更好。Traditional K-Means algorithm is sensitive to the initial centers and easy to get stuck at locally optimal value. To solve such problems, this paper presents an improved K-Means algorithm based on genetic algorithm. It combines the locally searching capability of the K-Means with the global optimization capability of genetic algorithm, and introduces the K-Means operation into the genetic algorithm of adaptive crossover probability and adaptive mutation probability, which overcomes the sensitivity to the initial start centers and locality of K-Means. Experimental results demonstrate that the algorithm has greater global searching capability and can get better clustering.

关 键 词：K均值算法 聚类中心 遗传算法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]