| 一种改进的SMO算法 |
| |
| 引用本文： | 张召 黄国兴 鲍钰. 一种改进的SMO算法[J]. 计算机科学, 2003, 30(8): 128-129 |
| |
| 作者姓名： | 张召 黄国兴 鲍钰 |
| |
| 作者单位： | 华东师范大学计算机系,上海,200062 |
| |
| 摘 要： | In this paper we have pointed out an important source of inefficiency in SMO algorithm that is caused by the operation with a single threshod value. We have suggested modifications of SMO algorithm that overcome the problem by efficiently maintaining and updating two threshold parameters. Our experiments show that these modifi-cations speed up the SMO algorithm.
|
| 关 键 词： | SMO算法 支持向量机 机器学习算法 目标函数 |
An Improved SMO Algorithm |
| |
| ZHANG Zhao HUANG Guo-Xing BAG Yu. An Improved SMO Algorithm[J]. Computer Science, 2003, 30(8): 128-129 |
| |
| Authors: | ZHANG Zhao HUANG Guo-Xing BAG Yu |
| |
| Abstract: | In this paper we have pointed out an important source of inefficiency in SMO algorithm that is caused by the operation with a single threshod value. We have suggested modifications of SMO algorithm that overcome the problem by efficiently maintaining and updating two threshold parameters. Our experiments show that these modifications speed up the SMO algorithm. |
| |
| Keywords: | SVM SMO KKT condition |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
点击此处可从《计算机科学》下载全文 |
