|
|||||
|
|
| 基于SMO的不同惩罚系数的SVM算法 | |
| 引用本文： | 王娟娟,任秋实. 基于SMO的不同惩罚系数的SVM算法[J]. 信息技术, 2006, 30(10): 45-47 |
| 作者姓名： | 王娟娟 任秋实 |
| 作者单位： | 上海交通大学生物医学工程系,上海,200030 |
| 摘 要： | 非平衡数据集的分类问题经常出现在许多实际应用中。支持向量机在处理这一类问题时，整体分类性能比较低。为此，Veropoulos提出的采用不同惩罚系数的改进算法可以较好的解决此类问题。此外，可以利用序列最小优化算法简单快速的解决上述优化问题。 |
| 关 键 词： | 支持向量机 序列最小优化 不平衡数据 |
| 文章编号： | 1009-2552(2006)10-0045-02 |
| 修稿时间： | 2006-04-05 |
SVM algorithm with different error costs based on SMO |
|
| WANG Juan-juan,REN Qiu-shi. SVM algorithm with different error costs based on SMO[J]. Information Technology, 2006, 30(10): 45-47 | |
| Authors: | WANG Juan-juan REN Qiu-shi |
| Abstract: | Unbalanced data classification often arises in many practical applications.When SVM is applied to the classification of unbalanced datasets,it leads to a poor classification performance.In order to overcome the difficulty,Veropoulos suggested using different error costs for different class,which can handle the problem of unbalanced data classification well.Besides,this paper applied sequential minimal optimization to solving the above optimization problem. |
| Keywords: | support vector machine sequential minimal optimization unbalanced dataset |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ | |