基于FCM的无监督最优模糊聚类算法  

Unsupervised optimal fuzzy clustering algorithm based on fuzzy c-means

作　　者：高永清[1] 陈志红[1] 黄鹤玲[1] 倪钊滨[1] 李霆[1] 

机构地区：[1]五邑大学信息学院,江门529020

出　　处：《信息技术》2009年第7期69-71,共3页Information Technology

摘　　要：基于模糊c-均值算法的无监督最优模糊聚类算法集合了模糊c均值算法与无监督最优聚类算法的优点,它通过逐渐改变聚类数c,依据一些有效性衡量尺度,能无监督搜索出最优聚类数c。通过对距离测量尺度的改进,使聚类不受类形状的影响,以达到具备更高准确率的聚类效果。仿真实验结果表明,新算法不仅能准确找出聚类数,而且跟单纯的模糊c均值算法比,具有更好的聚类效果。The unsupervised optimal fuzzy clustering algorithm based on the fuzzy c-means algorithm gathered the advantages of the fuzzy c-means algorithm and the unsupervised optimal Clustering algorithm. By changing the number of the clustering c gradually, basing on some measures of the effectiveness, the optimal clustering can be found out without supervising. By improving the measurement of the distance, the clustering can not be influenced by the shape of the class, in order to achieve a higher rate of correct clustering results. The simulation result showed that the new algorithm can not only find out the number of the clustering, but also has a better clustering effect compared to the fuzzy c-means algorithm.

关 键 词：模糊C均值算法 聚类有效性 无监督最优聚类 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]