FCM聚类算法与改进层次聚类算法的结合

模糊c-均值聚类算法(fuzzy C-means 简称FCM)和层次聚类算法是两种非常重要的聚类算法.由于FCM算法对初始聚类中心敏感,并且需要人为确定聚类类别数,这样收敛结果易陷入局部最优解.通过对这两种聚类算法的分析,首先对传统的凝聚层次聚类算法提出了改进,然后用改进的凝聚层次聚类算法得到最佳聚类数和初始聚类中心,最后用FCM算法进行再次聚类,以此得到更好的聚类结果并且减少了执行时间和迭代次数.

Combine of FCM Algorithm and Improved Hierarchical Clustering Algorithm

Fuzzy c-means algorithm(FCM)and hierarchical clustering algorithm are two important clustering algorithms.Because FCM is sensitive to the initial cluster centers and it need to determine the number of cluster type beforehand,so the result is easy to be local optimal.Based on the analysis of the two clustering algorithm,the hierarchical clustering algorithm firstly is improved,then it used the algorithm to get the number of cluster type and initial cluster centers,at last it uses FCM to perform the cluster.I...