摘要
在2类支持向量机(SVM)的基础上,综合分级聚类和决策树的思想构造多类支持向量机,并由此构建了中医舌像分类模型。该算法首先对训练集中距离最近的两类依次进行合并,得到一棵自底向上的决策树,再通过训练构造决策树的各级分类器,从而减少了分类器数量和支撑向量数量,加快了决策速度。由于采用了决策树的思想,从而避免了拒绝分类区和重复分类区的出现,提高了准确率。实验结果表明,与其他多类支持向量机方法相比,分级聚类SVM方法能够提高分类速度和准确率。
The basic Support Vector Machine（SVM） was designed for two-class problem.A new support vector machine based on hierarchical clustering and decision tree was proposed to solve the multi-class recognition problems such as tongue images classification.Given a training set,a decision tree was built by hierarchical clustering.The structure was simplified,the rate of identify was expedited,and the accuracy of identify was improved.The experiments based on tongue images show that the proposed algorithm is efficient to the classification of tongue images.
出处
《计算机应用》
CSCD
北大核心
2010年第12期272-273,276,共3页
journal of Computer Applications
关键词
决策树
分级聚类
舌像
分类
支持向量机
decision tree
hierarchical clustering
tongue image
classification
Support Vector Machine（SVM）
