以拟合优度为基础的两个分类算法及其在癌细胞自动识别中的应用
Two Classification Algorithms Based on the Coodness of Fit and Their Application in the Automated Recognition of Cancer Cells
-
摘要: 本文提出了集群方法"亦参数分级集群算法"及更为精确的拟合优度判据,并与传统的系统聚类方法进行了对比。将该方法与K一最近邻判决规则结合,提出了用于判别的固定邻域判决算法。全部算法用BCY语言在TQ-16机上实现。处理了细胞图象自动分析数据并取得了有益的结果。Abstract: In this paper, further discussion on the nonparametric clustering is made, An improved nonparametric classical clustering algorithm and a more accurate criterion based on the goodness of fit are advanced, Combining this method with the K-nearest neighbor decision rule, a fixed neighborhood decision algorithms are realized in the BCY languge on The TQ-16 computer. The cells are divided into classes by the above methods and useful results are achieved.