A Step-Based Deep Learning Approach for Network Intrusion Detection  

作　　者：Yanyan Zhang Xiangjin Ran 

机构地区：[1]Jilin Business and Technology College,Changchun,130507,China [2]College of Earth Sciences,Jilin University,Changchun,130061,China

出　　处：《Computer Modeling in Engineering & Sciences》2021年第9期1231-1245,共15页工程与科学中的计算机建模（英文）

基　　金：This work was supported by the Education Department of Jilin Province(No.JJKH20180518KJ);Science and Technology Research Project of Jilin Business and Technology College(No.kz2018002).

摘　　要：In the network security field,the network intrusion detection system(NIDS)is considered one of the critical issues in the detection accuracy andmissed detection rate.In this paper,amethod of two-step network intrusion detection on the basis of GoogLeNet Inception and deep convolutional neural networks(CNNs)models is proposed.The proposed method used the GoogLeNet Inception model to identify the network packets’binary problem.Subsequently,the characteristics of the packets’raw data and the traffic features are extracted.The CNNs model is also used to identify the multiclass intrusions by the network packets’features.In the experimental results,the proposed method shows an improvement in the identification accuracy,where it achieves up to 99.63%.In addition,the missed detection rate is reduced to be 0.1%.The results prove the high performance of the proposed method in enhancing the NIDS’s reliability.

关 键 词：Network intrusion detection system deep convolutional neural networks GoogLeNet Inception model step-based intrusion detection 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]