改进的ReliefF算法在入侵检测中的应用

Application of Improvement of ReliefF Algorithm in Intrusion Detection

针对入侵检测领域计算复杂、时间复杂度高的难题,本文通过添加入侵特征间的相关度量准则,进一步改进Re-ReliefF入侵特征选择方法,增强区分特征间互信息的能力,达到更有效降维。目标是在改进ReReliefF算法的基础上,实现更佳的入侵检测效果。实验采用KDD CUP 99数据集,对数据的41维特征进行选择,采用支持向量机作为分类器,结果表明所提出的改进方法在分类的耗时和准确率略好的情况下,提高27%的误报率。

In view of the problem of high computational complexity and time complexity from the intrusion detection domain,this article by adding the related measurement standards and the invasion characteristics to further improve the Re-Relief intrusion feature selection method,to strengthen the ability of distinguishing characteristics between mutual information,and to achieve more effective dimensionality reduction. Target is based on improving the Re-Relief algorithm to achieve better intrusion detection. The experiment uses KDD CUP 99 data sets to select 41-dimensional feature data by supporting of vector machine（ svm） as classifier. It turned out that the improved methods who was proposed in the case of the time-consuming and classification accuracy slightly better situation,and 27% increase in the rate of false positives.