| 基于SVM的入侵检测系统中特征权重优选方法综述 |
| |
| 引用本文： | 解永娟,王汝传,任勋益.基于SVM的入侵检测系统中特征权重优选方法综述[J].信息安全与通信保密,2007(8):126-129. |
| |
| 作者姓名： | 解永娟 王汝传 任勋益 |
| |
| 作者单位： | 1. 南京邮电大学,计算机学院,江苏,南京,210003
2. 南京邮电大学,计算机学院,江苏,南京,210003;南京大学,计算机软件新技术国家重点实验室,江苏,南京,210093 |
| |
| 基金项目： | 国家自然科学基金;江苏省自然科学基金;江苏省高技术研究发展计划项目;国家高技术研究发展计划(863计划);江苏省南京市高科技项目;国家重点实验室基金;江苏省重点实验室基金 |
| |
| 摘 要： | 基于统计学习理论的支持向量机有较好的泛化能力,然而当样本含有与该问题不完全相关甚至完全无关的特征时,会使得各个特征对问题的相关程度差异很大,为了提高分类的正确率,对各个特征进行加权尤为重要。在入侵检测系统中,网络中的特征对分类结果的影响程度也是不同的,本文列举了对这些特征进行加权的几种方法。
|
| 关 键 词： | 特征加权 支持向量机 入侵检测 |
| 文章编号： | 1002-0802(2007)08-0126-04 |
| 修稿时间： | 2007年6月15日 |
Methods of Feature Weights Optimization in Intrusion Detection System based Support vector machine |
| |
| XIE Yong-juan,WANG Ru-chuan,REN Xun-yi.Methods of Feature Weights Optimization in Intrusion Detection System based Support vector machine[J].China Information Security,2007(8):126-129. |
| |
| Authors: | XIE Yong-juan WANG Ru-chuan REN Xun-yi |
| |
| Abstract: | Support vector machines based on the theory of statistical learning have a good generalization performance. However, if the sample contains features that are not completely relative or even completely irrelative to the problem, it would make the difference of correlative degree for various features to the problem large. The feature weighting is particularly important to improve the accuracy of classification. In IDS, the impact of network features on classification is different. The paper lists several methods for the feature weighting. |
| |
| Keywords: | feature weighting support vector machine intrusion detection |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
