计算机技术与发展Issue(7)：181-184,4.DOI:10.3969/j.issn.1673-629X.2014.07.045
改进的二叉树支持向量机在多分类中的应用
Application of Improved Binary Tree Support Vector Machine in Multi-classification
摘要
Abstract
The quality of algorithms has directly impacted on the final classification results in multi-class classification. In current algo-rithms for multi-classification,those which are based on Support Vector Machine ( SVM) have better comprehensive performance than others. But they also face some common problems,such as unclassifiable regions and low efficiency. For these problems,a modified bina-ry tree SVM multi-classification algorithm which is based on the effect of distance and distribution of classes to inter-class separability is proposed,using the strategy of the easiest class to separate for the first partition to establish the structure of the tree. Tests in different data sets show that this method can not only solve the unclassifiable regions,but also can improve the efficiency and accuracy of classification.关键词
支持向量机/多分类/二叉树/超球体Key words
support vector machine/multi-classification/binary tree/hypersphere分类
信息技术与安全科学引用本文复制引用
李燕玲,苏一丹..改进的二叉树支持向量机在多分类中的应用[J].计算机技术与发展,2014,(7):181-184,4.基金项目
教育部人文社会科学研究项目(11YJAZH080) （11YJAZH080）
