一种基于类间距阈值的模糊聚类算法  

A FUZZY CLUSTERING ALGORITHM BASED ON THE THRESHOLD OF CLUSTERS' INTERVAL

作　　者：谢宇[1] 程维明[1] 

机构地区：[1]上海大学机电工程与自动化学院,上海200072

出　　处：《计算机应用与软件》2008年第9期248-249,共2页Computer Applications and Software

摘　　要：针对模糊C-均值(Fuzzy C-Means,FCM)算法需要事先设定聚类数C,若设定不当,容易导致错分类的问题,提出了一种利用类间距阈值搜索聚类数的方法来确定最佳聚类数。算法可以自适应确定最佳聚类数,仿真结果表明了利用类间距阈值方法确定聚类数的有效性。Fuzzy C-Means is one of the clustering algorithms based on the optimization of an objective function. If the cluster number is not defined appropriately wrong classification may be induced. A new algorithm based on the threshold of clusters＇ distance is proposed. The new algorithm can detect the optimal cluster number automatically. The computer simulation resuhs show that the new algorithm is effective.

关 键 词：类间距 模糊聚类 FCM 

分 类 号：TP274.2[自动化与计算机技术—检测技术与自动化装置] TU113.3[自动化与计算机技术—控制科学与工程]