Abstract: This paper presents one clustering discrimination method which is different from the previous methods.It starts with the categories，constructs core-sets for clustering discrimination，and establishes a model to it.This paper uses the smallest circle as the core-sets for clustering discrimination in the 2-dimensional space and does numerical experiment on the diagnosis of breast cancer.In this model，this paper first selects the indicators，processes the data，and then takes the smallest circle as core-sets to establish the model，at last tests another 69 data.The rate of the misjudgment is 4.35%.

Key words: core-sets, centre of clustering, clustering discrimination

摘要： 区别于传统的聚类方法，提出了以类为起点，通过构造闭包进行聚类的新方法，并建立了聚类判别模型，此模型给出了对于闭包间的交叉区域的检验点的判别准则。然后针对二维的聚类问题，提出了以最小圆为闭包的聚类判别模型，并对乳房肿瘤病例进行数值实验。对于乳房肿瘤病例，首先进行了指标选取、数据预处理，然后以最小圆为闭包建立了模型，最后对69个待检测数据进行检验，结果误判率为4.35％。

关键词: 闭包, 聚类中心, 聚类判别