论文部分内容阅读
为解决入侵检测领域计算复杂度、时间复杂度高的难题,达到更优秀的入侵检测效果,有效降维,在原有的ReFCBF算法的基础上,提出增强区分特征间互信息的能力,以在改进的Re-ReliefF算法的基础上,实现更佳的入侵检测效果为目标。实验采用DARPA 2000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,实验结果表明,该改进方法在分类的耗时和误报率略好的情况下,提高了30%的准确率。
In order to solve the problem of computational complexity and time complexity in the field of intrusion detection, achieve better intrusion detection effect and reduce dimension effectively. Based on the original ReFCBF algorithm, it is proposed to enhance the ability of distinguishing mutual information between features, Re-ReliefF algorithm based on the realization of better intrusion detection results as the goal. The experiment uses DARPA 2000 data set to select the 41-dimensional features of the data and uses SVM as the classifier. Experimental results show that the proposed method improves the classification accuracy by 30% The accuracy rate.