【摘要】 模糊C-均值(FCM)聚类算法是目前最流行的数据集模糊划分方法之一.但是,有关聚类类别数的合理选择和确定,即聚类有效性分析,对FCM算法而言仍是一个开放性问题.为此,本文结合数据集的几何结构信息和FCM算法的模糊划分信息,重新定义了划分矩阵,进而利用划分模糊度提出了一种新的模糊聚类有效性函数.实验结果表明该方法是有效的且具有良好的鲁棒性.更多还原

【Abstract】 Fuzzy c-mean (FCM) is one of popular and effective algorithms for data partition. However, the reasonable determination of the cluster number, i. e. clustering validity is still an open problem for FCM. For this purpose, by combining the geometrical structure of the data set with its fuzzy partition information, this paper redefines the partition matrix of the data set, and further presents a clustering validity function based on the partition fuzzy degree. Experimental results illustrate the effectiveness and robustness of the proposed clustering validity function.更多还原