聚类是非监督学习的关键问题.本文在模糊最小-最大聚类网络(FMMCN)和分层聚类思想的基础上,提出一种分层模糊最小-最大聚类算法.与目前的常用聚类算法相比,该方法可以根据问题需要动态确定聚类数目,并克服 FMMCN 样本输入次序依赖性的缺陷.对相关数据集的实验结果表明该方法具有优良的聚类性能. Clustering is the key issue of unsupervised learning.This paper proposes a hierarchical fuzzy minimum-maximum clustering algorithm based on the idea of ​​fuzzy minimum-maximum clustering network (FMMCN) and hierarchical clustering.Compared with the current common clustering Compared with the proposed algorithm, this method can dynamically determine the number of clusters according to the needs of the problem and overcome the shortcomings of the input order dependency of FMMCN samples.Experimental results on the relevant data sets show that the proposed method has excellent clustering performance.