机构地区: 华南理工大学计算机科学与工程学院
出 处: 《微电子学与计算机》 2007年第6期156-157,161,共3页
摘 要: 给出了一种新的基于FCM的聚类算法,能根据由数据的分布的特性自动获取要聚类的数目,在新增数据后,可以进行增量式聚类,结果对孤立点不敏感,并能完成FCM不具备的非球型或椭球型分布的数据集的聚类,实验结果显示算法的有效性和优越性。 A new clustering algorithm based on FCM is presented. It can obtain the number automatically according to the distribution characteristic of data set, and it can realize incremental clustering, the result is not sensitive to outlier, it can accomplish clustering of not spherical or elliptical data set that the FCM don't possess. Experimental results show the new algorithm is effective and advantageous.
领 域: [自动化与计算机技术] [自动化与计算机技术]