基于KNN算法及禁忌搜索算法的特征选择方法在入侵检测中的应用研究  

A Research and Application of Feature Selection Based on KNN and Tabu Search Algorithm in the Intrusion Detection

作　　者：张昊[1] 陶然[1] 李志勇[1] 蔡镇河[1] 

机构地区：[1]北京理工大学信息科学技术学院,北京100081

出　　处：《电子学报》2009年第7期1628-1632,共5页Acta Electronica Sinica

基　　金：国防基础科研项目(No.C1120060497-06)

摘　　要：在入侵检测中应用特征选择能够在保持原有信息完整性的基础上,去除其中的冗余特征,有效地提高入侵检测系统的检测速度.本文提出了一种新的特征选择方法,即基于KNN算法及禁忌搜索算法的特征选择方法.实验结果表明该特征选择方法能够有效去除网络数据信息中的冗余特征,减少特征选择时间;并且能够在保证检测准确率的前提下,有效提高系统的检测速度.Utilizing the feature selection in the intrusion detection can delete the redundant features on the base of protecting the integrity of original data and improve the detection speed of the system efficiently. This paper proposes a new feature selection method that is based on KNN and Tabu search algorithm. The experiment result shows that this method can remove the redundant features, and reduce the time of feature selection. This method not only can guarantee the correct rate of detection but also improve the detection speed efficiently.

关 键 词：入侵检测 特征选择 特征关联性 禁忌搜索 

分 类 号：TN915.08[电子电信—通信与信息系统]