深度置信网络结合递归特征添加的网络入侵检测方法  

NETWORK INTRUSION DETECTION METHOD BASED ON DEEP BELIEF NETWORK AND RECURSIVE FEATURE ADDITION

作　　者：赵荷[1] 盖玲[2] Zhao He;Gai Ling(Department of Computer Science and Engineering,Chengdu Neusoft University,Chengdu 611844,Sichuan,China;School of Management,Shanghai University,Shanghai 200444,China)

机构地区：[1]成都东软学院计算机科学与工程系,四川成都611844 [2]上海大学管理学院,上海200444

出　　处：《计算机应用与软件》2020年第11期304-310,共7页Computer Applications and Software

基　　金：四川省科技厅重大科技专项项目(18ZDZX0078)。

摘　　要：针对互联网零日攻击严重威胁网络安全的问题,提出一种深度置信网络(Deep Belief Network,DBN)结合递归特征添加(Recursive Feature Addition,RFA)的网络入侵检测方法。采用深度置信网络对网络入侵特征进行提取,并基于二元组编码技术将长字符串的特征转化为二进制编码;使用递归特征添加方法对影响网络入侵检测性能的主要特征进行选择。为获取更好的入侵检测性能,提出综合考虑检测准确率、检出率和误报率的入侵检测性能评估函数,从而有效改善抵御互联网攻击的能力。实验结果表明,相较于K-最邻近(K-Nearest neighbor,KNN)算法等传统的入侵检测算法,该算法的检测准确率提升8%以上,保证了互联网的安全性。To solve the problem that zero-day attacks on the Internet which seriously threaten the network security,we propose a network intrusion detection method based on deep belief network(DBN)and recursive feature addition(RFA).The DBN was used to extract the network intrusion features,and the binary encoding technology was used to transform the long-string-features into binary encodings.Then,the RFA was used to select the main features affecting network intrusion detection performance.To obtain better intrusion detection performance,we proposed an intrusion detection performance evaluation function considering detection accuracy,detection rate and false alarm rate,which effectively improve the ability to resist the Internet attacks.The experimental results show that compared with traditional intrusion detection algorithms such as K-nearest neighbor(KNN)algorithm,our algorithm is improved by more than 8%,thus ensuring the security of the Internet.

关 键 词：深度学习 递归特征添加 网络入侵检测 互联网攻击 网络安全 零日攻击 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]