| 一种新的SVM多类分类算法 |
| |
| 引用本文： | 霍颖瑜,王晓峰. 一种新的SVM多类分类算法[J]. 佳木斯大学学报, 2006, 24(4): 476-478 |
| |
| 作者姓名： | 霍颖瑜 王晓峰 |
| |
| 作者单位： | 广东佛山科学技术学院理学院,广东,佛山,582000;广东北电研发中心,广东,广州,510200 |
| |
| 摘 要： | 给出了一种基于编码二叉树的支持向量机(SVM,Support Vector Machine)的多类分类算法.首先,定义了一种构造编码二叉树的方法,在此基础上合理的使用每个训练样本对应的编码来对多类样本进行划分,使之转化为两类分类问题.由算法的实现过程可以看出,本算法可以大大减少子分类器的构造个数,从而简化了多类SVM分类算法.
|
| 关 键 词： | 编码二叉树 SVM 多类分类 |
| 文章编号： | 1008-1402（2006）04-0476-03 |
| 收稿时间： | 2006-07-21 |
| 修稿时间： | 2006-07-21 |
A New Algorithm for Multi-category Support Vector Machine |
| |
| HUO Ying-yu,WANG Xiao-feng. A New Algorithm for Multi-category Support Vector Machine[J]. Journal of Jiamusi University(Natural Science Edition), 2006, 24(4): 476-478 |
| |
| Authors: | HUO Ying-yu WANG Xiao-feng |
| |
| Affiliation: | 1. Foshan University, Foshan 582000, China;2. Guang dong Nortel Tele Communications Equipment Co Ltd,Guangzhong 510220,China |
| |
| Abstract: | This paper presents an algorithm of SVM(Support Vector Machine) for multi-classification.Firstly a method of constructing bintree-coding is defined.Then the multi-classes samples are divided into two classes according to the code corresponding to each training sample.The algorithm can decrease the number of sub-classification greatly and simplify the SVM algorithm for multi-classification. |
| |
| Keywords: | SVM |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
