基于深度学习的网络入侵检测研究综述  

Survey of Network Intrusion Detection Based on Deep Learning

作　　者：肖建平 龙春[1,2] 赵静 魏金侠[1] 胡安磊 杜冠瑶[1,2] XIAO Jianping;LONG Chun;ZHAO Jing;WEI Jinxia;HU Anlei;DU Guanyao(Computer Network Information Center,Chinese Academy of Sciences,Beijing 100190,China;School of Computer Science and Technology,University of Chinese Academy of Sciences,Beijing 101408,China;China Internet Network Information Center,Beijing 100190,China)

机构地区：[1]中国科学院计算机网络信息中心,北京100190 [2]中国科学院大学,计算机科学与技术学院,北京101408 [3]中国互联网络信息中心,北京100190

出　　处：《数据与计算发展前沿》2021年第3期59-74,共16页Frontiers of Data & Computing

基　　金：中国科学院“十四五”网信专项先期建设项目(WX145XQ10,WX145XQ11)。

摘　　要：【目的】互联网的迅速发展给人们的生活带来了极大的便利,然而各种网络攻击行为也日益增加,网络空间面临着严重的威胁。入侵检测在防护网络攻击中发挥着关键作用。【文献范围】近年来,深度学习方法在入侵检测领域得到了广泛应用。本文通过广泛的文献调查,选取了该领域的最新研究工作。【方法】首先介绍了当前的网络安全形势,并总结了入侵检测系统的类型、数据集和评估方法,然后在检测技术层面,论述了基于传统机器学习方法的入侵检测和基于深度学习的入侵检测。最后,对入侵检测技术未来的研究方向进行了展望。【结果】通过分析对比,基于深度学习方法的入侵检测系统通常具有更好的性能。【局限】受限于获取文献的范围,没有对基于深度学习的入侵检测方法所解决的问题进行对比。【结论】基于深度学习方法的入侵检测技术在处理高维数据、获取数据中隐藏信息、解决网络中数据不平衡问题等方面具有优势,未来在入侵检测领域会应用地越来越广泛。[Objective]The rapid development of the Internet has brought great convenience to people's life.However,various malicious network attacks are also increasing,and cyberspace is facing serious threats.Intrusion detection plays a key role in preventing network attacks.[Coverage]In recent years,deep learning methods have been widely used in the field of intrusion detection.In this paper,through an extensive literature survey,we select the latest research work in this field.[Methods]Firstly,this paper introduces the current network security situation and summarizes the types,data sets,and evaluation methods of intrusion detection systems.In the aspect of detection technology,it discusses traditional machine learning and deep learning methods.Finally,it introduces the future research direction of intrusion detection technology.[Results]Through analysis and comparison,it shows that intrusion detection systems based on deep learning methods usually have better performance.[Limitations]Due to the scope of the available literature,this article does not make a comparison in the view of the problems solved by various intrusion detection methods based on deep learning.[Conclusions]Intrusion detection technologies based on deep learning have advantages in processing high-dimensional data,obtaining hidden information in data,and solving the problem of data imbalance in the network.In the future,it will be more and more widely used in the field of intrusion detection.

关 键 词：网络安全 入侵检测 深度学习 机器学习 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]