基于人进行聚类判断所遵循的基本原则,即聚类中对象间距离应小于聚类间距离,本文提出了一种聚类学习的新方法。该方法无需用户事先给定聚类个数K,且其算法复杂度基本为O(nm)。此外,该聚类学习新方法,在处理大规模(几十万至几百万个)数据对象时,也具有良好的工作效能。有关实验测试结果,充分说明了该聚类学习新方法的有效性。 Based on the basic principle that human beings judge by clustering, that is, the distance between objects in the cluster should be less than the distance between the clusters. In this paper, a new clustering learning method is proposed. The method does not require the user to give the number of clusters K in advance, and the algorithm complexity is basically O (nm). In addition, this new method of clustering learning also works well when dealing with large-scale (hundreds of thousands to millions) data objects. The experimental results show that the new clustering learning method is effective.