具有约束条件的C分类

本文给出了一种在类别数一定且各类中样本数一定的条件下的两种分类方法,它将传统分类方法向类中心聚集改变为向类优向量聚集,使具有此种约束条件的Ｃ分类更加合理。

C-Clustering with Restricted Conditions

In this paper two kind of C-clustering algorithms are proposed in which the number of clusters and the number of sample points in every cluster is known. It changes the thoughts from each class clustering to the center of the cluster into the each class clustering to the optimal vevter of the cluster. It may be reasonable to expect "good"classification under this restricted conditions, A practical example are used to illustrate the result.