针对传统快速k-近邻分类算法的缺陷,提出了一种基于近邻搜索的快速k-近邻分类算法———超球搜 索法。该方法通过对特征空间的预组织,使分类在以待分样本为中心的超球内进行,有效地缩小了搜索范围。 -Rapid response to traditional k-neighbors of the defect classification algorithm, a fast search based on neighbor k-neighbor classification algorithm

中括号内的序号为寻找聚类中心的迭代运算的次序号。聚类中心的向量值可任意设定，例如可选开始的K个模式样本的向量值作为初始聚类中心。-The number in brackets in order to find the cluster centers iteration sequence number. Cluster center value can be set to, for example, the beginning of an optional mode of samples to t

k mean cluster for radial basis function neural network