论文部分内容阅读
提出使用粗糙集分类(RSC)算法进行智能化的网络入侵检测.该方法可以在生成检测规则之前完成特征排序,且不需要多次重复迭代计算,提高了入侵检测系统的效率;同时,生成的检测规则是“if-then”格式的产生式,易于解释.仿真实验表明,RSC对Probe和DoS攻击具有比支持向量机(SVM)略好的高检测率,但是训练时间比SVM更长,采用混杂遗传算法求解粗糙集约简可进一步减少RSC的训练时间.
This paper proposes an intelligent network intrusion detection using rough set classification (RSC) algorithm.This method can finish the feature ordering before the generation of detection rules, and does not need repeated iterative computation, which improves the efficiency of intrusion detection system. At the same time, The test rules are easy to interpret in the format of “if-then”. The simulation results show that RSC has a slightly better detection rate than Probe SVM for Probe and DoS attacks, but the training time is longer than SVM , The use of hybrid genetic algorithm to solve rough set reduction can further reduce the RSC training time.