基于样本密度信息与竞争网络的聚类中心点获取算法  

Access to Center Cluster Algorithm Based on Sample Density of Information and Competition Network

作　　者：王磊 杜文莉[1] 祁荣宾[1] 钱锋[1] 

机构地区：[1]华东理工大学化工过程先进控制和优化技术教育部重点实验室,上海200237

出　　处：《华东理工大学学报（自然科学版）》2009年第4期648-654,共7页Journal of East China University of Science and Technology

基　　金：国家杰出青年科学基金(60625302);国家自然科学基金面上项目(20876044);上海市科技启明星项目(08QA14021);高等学校博士学科点专项科研基金新教师基金项目(200802511011);上海市重点学科建设项目资助(B504)

摘　　要：在聚类分析中初值的选取对聚类结果起着关键性的作用。本文在Chiu算法思想的基础上,提出了一种根据样本密度信息获取中心点的算法。该算法不需要任何参数的设定就可实现中心点的获取;之后再通过竞争网络对获取到的中心点进行训练,使中心点更加靠近每一类的中心。仿真实验表明:该算法是有效的且具有很高的可靠性,保证了网络训练前的中心点分布在不同的类簇中,提高了网络的训练效率。In cluster analysis, the selection of the initial central point plays a crucial role on the results of the cluster. By using the sample density of information, this paper proposes an algorithm to obtain the central point. This algorithm can obtain the central point without setting any parameters. And then, by training the obtained central point via competition network, the central point can be made closer to the center of each cluster. The simulation results show that this algorithm is effective and high reliability, and can ensure that the central point to be located in a different type of cluster before the training of competition network, which can improve the efficiency of the network.

关 键 词：聚类分析 中心点 密度函数 竞争网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]