基于遗传算法的聚类分析  

Cluster Analysis based on Genetic Algorithm

作　　者：苏良昱[1] 苏良碧[2] 

机构地区：[1]许昌学院电气信息工程学院,河南许昌461000 [2]内蒙古大学电子信息工程学院,呼和浩特010021

出　　处：《电脑开发与应用》2011年第2期7-9,共3页Computer Development & Applications

摘　　要：遗传算法是一种模拟自然进化的优化搜索算法,它仅依靠适应度函数就可以搜索最优解。介绍了一种基于遗传算法的聚类分析方法,采用浮点数编码方式对聚类的中心进行编码,并用特征向量与相应聚类中心的欧氏距离的和来判断聚类划分的质量,通过选择、交叉和变异操作对聚类中心的编码进行优化,得到使聚类划分效果最好的聚类中心。实验结果显示,该方法的聚类划分能得到比较满意的效果。Genetic Algorithm is a optimized search approach which simulates the natural evolution,it can search the optimal solution only by fitness function.This paper introduce a cluster analysis algorithm which base on GA,it uses floating-point encoding method to encode the center of the cluster,and use Euclidean distance between the feature vector and the corresponding cluster center to determine the quality of clustering,it optimizes the encoding of the cluster centers by selection,crossover and mutation operations,it can obtain the cluster centers which can make the best clustering.Experimental results show that clustering use this method can reach satisfied results.

关 键 词：遗传算法 聚类 适应度函数 GA 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]