基于改进变分自动编码器的入侵检测模型构建及仿真  

Construction and simulation of intrusion detection model based on improved COD-VAE model

作　　者：曹春梅 CAO Chunmei(Information Technology Center,Hefei Preschool Teachers College,Hefei 230013,China)

机构地区：[1]合肥幼儿师范高等专科学校信息技术中心,安徽合肥230013

出　　处：《河南工程学院学报（自然科学版）》2023年第2期63-69,80,共8页Journal of Henan University of Engineering：Natural Science Edition

基　　金：安徽省高等学校省级质量工程项目(2021zyyh034)。

摘　　要：针对现有网络流量入侵检测查准率低的问题,提出了一种基于改进变分自动编码器的入侵检测方法,先在变分自动编码器上增加判别器实现网络流量的入侵检测,之后在CICIDS2017数据集上对所提方法进行验证。结果表明,所提方法对正常流量与异常流量检测的平均查准率、召回率、F1值均达到87%以上,且平均AUC值达90%。相较于CE-SAE模型和传统变分自动编码器,所提方法在各项指标上的表现更好,具有明显优势。Aiming at the problem of low accuracy of intrusion detection in existing network traffic,an intrusion detection method based on improved COD-VAE model is proposed.Methods by adding a discriminator to the COD-VAE model and using the GAN network confrontation learning method to build the model,the network traffic intrusion detection is realized.Finally,the proposed method is verified on CICIDS2017 data set.The results show that the proposed intrusion detection method based on the improved COD-VAE model has good detection performance.The average precision,recall and F1 value of normal and abnormal traffic are more than 87%,and the average AUC value is 90%.Compared with CE-SAE model and traditional COD-VAE model,the proposed method performs better in various indicators and has obvious advantages.

关 键 词：网络安全 入侵检测 生成对抗网络 变分自动编码器 查准率 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]