| SVM入侵检测系统中数据预处理方法改进 |
| |
| 引用本文： | 李秦渝.SVM入侵检测系统中数据预处理方法改进[J].交通科技与经济,2009,11(3):94-95. |
| |
| 作者姓名： | 李秦渝 |
| |
| 作者单位： | 兰州交通大学,自动化及电气工程学院,甘肃,兰州,730070;兰州城市学院,信息工程学院,甘肃,兰州,730070 |
| |
| 摘 要： | 由于网络数据的复杂性和不规范性制约着SVM分类器的精度，当前被广泛使用的数据预处理方法显得过于单一。因此，提出一种改进的数据预处理方法。首先，利用异构数据集上的奇异距离函数HVDM对数据进行归一化处理；然后，使用最近邻法对数据集修剪得到最后的训练样本，并且通过实验证明该方法可以提高SVM分类器的精度。
|
| 关 键 词： | 数据预处理 入侵检测系统 SVM 改进 |
The Improvement to Data Pre-Processing Methods of Intrusion Detection System Based on SVM |
| |
| LI Qin-yu.The Improvement to Data Pre-Processing Methods of Intrusion Detection System Based on SVM[J].Technology & Economy in Areas of Communications,2009,11(3):94-95. |
| |
| Authors: | LI Qin-yu |
| |
| Institution: | 1. Lanzhou Jiaotong University;Automation and Electrical Engineering College;Lanzhou 730070;China;2.Lanzhou City University;Information Engineering College;China |
| |
| Abstract: | Because of the complexity and the non-normative of the data network restricted the accuracy of the SVM classifier to a large extent,and the current widely-used data pre-processing methods seem to be too single. An improved data pre-processing method was made to address this issue. Singular distance function HVDM of the heterogeneous data sets is used to normalize data at first, and then use the neaest neighbor method to cut the data set to get final training samples, and through an experiments to prove this... |
| |
| Keywords: | SVM |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
