摘要
二叉树支持向量机分类算法主要是构造一个偏二叉树或是构造一颗完全二叉树,但是偏二叉树分类的准确性虽高而分类的效率低,完全二叉树分类的效率高但是准确性不高。本文提出一种算法,结合了以上两种二叉树构造方法的优点,并且更能反映样本的真实分布。实验结果表明,新算法具有较高的推广性能。
Binary tree support vector machine is mainly to construct a classification algorithm partial binary tree or tectonic a single fully binary tree. But slant binary tree classification accuracy is high but low efficiency, fully binary tree classification is high efficiency but not high accuracy. This paper proposed an algorithm that combines the above two kinds of binary tree structure method＇s metrics and it better reflects the real distribution. Experimental results show that the new algorithm has good generalization performance.
出处
《微型机与应用》
2011年第6期14-16,19,21,共5页
Microcomputer & Its Applications
基金
安徽省教育厅重点项目(KJ2009A57)
关键词
二叉树
支持向量机
多类多分
球结构
binary tree
support vector machine
many kinds of points
ball structure
