基于FCM的决策树分类器设计
首发时间:2010-11-26
摘要:模糊C均值聚类(FCM)属于非监督分类,对于分类出的结果抑或是在分类结果上提取出的结果都不能提取分类的规则。决策树属于监督分类,是从一组无序、无规律的事例中归纳学习产生规则。本文利用隶属度,将FCM的结果提取出来以供决策树训练集的选取,这样就舍弃掉了分类规则里的不确定信息,从而提高了分类的精度,最后用MATLAB实验,对比分析了这种混合分类器的可行性。
For information in English, please click here
The Design of Decision Tree Based on FCM
Abstract:FCM is non-supervised classification, and the results or the abstracted classified can't be extracted the classification rules, decision tree is supervised classification, which can learn and obtain rules from groups of the instance in the disorder. In this paper, the results extracted by using degree of membership are used to selection of training set of decision tree. In this way, uncertain information can be abandoned, so it improves the accuracy of classification. Finally, it is proved that this way is feasible with MATLAB.
Keywords: FCM decision tree degree of membership MATLAB
论文图表:
引用
No.4392286513556129****
同行评议
共计0人参与
勘误表
基于FCM的决策树分类器设计
评论
全部评论0/1000