论文部分内容阅读
聚类算法在数据分析与图象处理等许多方面应用十分广泛,尤其是模糊C均值(FCM)聚类算法受到人们的普遍重视。象其它聚类算法一样,进行FCM聚类时,需事先确定一些参数,如:聚类类别数C模糊加权指数m、向量范数等。如何确定数据的最佳分类,使之能准确真实地反映实际数据的内部结构,这就是聚类的有效性问题。本文在实验的基础上对FCM聚类算法进行有效性分析,并提出了一个能表征FCM聚类有效性的启发性函数,得到了一些有用结论。
Clustering algorithms are widely used in many aspects such as data analysis and image processing. Especially, fuzzy C-means clustering (FCM) clustering algorithms are widely regarded by people. Like other clustering algorithms, some parameters need to be determined in advance when performing FCM clustering, such as clustering number C fuzzy weighted index m, vector norm and so on. How to determine the best classification of data, so that it can accurately reflect the actual structure of the internal data, which is the validity of clustering problem. In this paper, the validity of FCM clustering algorithm is analyzed on the basis of experiments, and a heuristic function that can characterize the validity of FCM clustering is proposed. Some useful conclusions are obtained.