一种基于参考点的快速k-均值算法

A Fast K-mean Algorithm Based on Reference and Density

聚类分析是模式识别的一个重要分支,以核心点和k-均值算法为基础,提出了一种基于参考点的快速k-均值算法;本算法以参考点作为第一个初始聚类中心,剩余初始聚类中心在核心点中选取,使得初始聚类中心能更好的反映模式样本集的几何特征,并且能减少迭代次数。

Clustering analysis is an important branch of pattern recognition, in this paper, based on the corepoint and k-mean algorithm, a fast k-mean algorithm based on reference and density is proposed, this algorithm regards reference point as the first initial clustering center, the remaining initial clustering center is selected from the core points so that the initial clustering center can better respond geometric features of pattern sample set and can reduce the number of iterations.