摘要
为了解决在真实网络环境中,异常数据比正常数据更难获得的问题,提出基于生成对抗网络的网络入侵检测系统GAN-NIDS。在训练阶段只使用正常数据,通过卷积操作压缩数据,使网络结构记住正常数据的深度特征。在测试时,正常数据通过生成器生成的数据与原始数据之间的损失(loss),远小于异常数据通过生成器生成的数据与原始数据之间的loss。在KDD99数据集上进行了试验,结果表明,相较于传统机器学习与一些深度学习,本系统有较好的检测效果。
In order to solve the problem that abnormal data is more difficult to obtain than normal data in a real network environment,a network intrusion detection system GAN-NIDS based on generative adversarial networks is proposed.In the training phase,only normal data is used,and the data is compressed through convolution operation,so that the network structure remembers the deep features of normal data.In the test,the loss between the normal data generated by the generator and the original data is much smaller than the loss between the abnormal data generated by the generator and the original data.On the KDD99 data set,the results show that compared with traditional machine learning and some deep learning,this system has better detection results.
作者
陈阳
王怀彬
CHEN Yang;WANG Huaibin(School of Computer Science and Engineering,Tianjin University of Technology,Tianjin 300384,China)
出处
《天津理工大学学报》
2021年第3期25-29,共5页
Journal of Tianjin University of Technology
