摘要
网络入侵数据集中存在的大量冗余和噪声特征严重影响检测系统的性能。针对该问题,提出一种基于Fisher分和支持向量机的入侵特征选择算法。通过对各维特征的Fisher分值排序,结合支持向量机分类算法,建立特征分类模型,筛选出具有最高检测率与误码率比值的最优特征组合。仿真结果表明,该算法筛选出的特征组合具有较高的检测率和较低的误码率,有效降低了检测系统的建模时间和测试时间,提高了系统性能。
There are many redundant and noisy characteristics in network intrusion detection data set,which leads to a bad performance of the detection system.To solve the problems,an intrusion feature selection algorithm based on Fisher value and support vector machine was proposed.By combining the support vector machine and sorting Fisher value of each dimensional feature,the optimal feature subset which owned the highest ratio of detection rate and bit error rate was selected.The simulation test results show that the method can eliminate noisy characteristics of the intrusion data set,reduce the modeling time and the testing time of the detection system,and improve the performance of the system.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第12期4145-4148,4190,共5页
Computer Engineering and Design
基金
广西自然科学基金项目(2012GXNSFAA053224)
保密通信重点实验室基金项目(9140C110404110C1106)
广西教育厅基金项目(201010LX156
CD10066X)
关键词
入侵检测
Fisher分
支持向量机
特征选择
数据标准化
intrusion detection
Fisher value
support vector machine
feature selection
data standardization
