基于卷积神经网络的入侵检测算法  

Intrusion Detection Algorithm Based on Convolutional Neural Network

作　　者：贾凡[1] 孔令智[1] 

机构地区：[1]北京交通大学电子信息工程学院,北京100044

出　　处：《北京理工大学学报》2017年第12期1271-1275,共5页Transactions of Beijing Institute of Technology

基　　金：国家自然科学基金资助项目(61374120;61673387);陕西省自然科学基金资助项目(2016JM6015)

摘　　要：作为深度学习的一种有效算法,深度卷积网络已成功应用在处理图像、视频和音频等领域.通过建立一卷积神经网络模型并应用于网络入侵检测,选取的卷积核与数据进行卷积操作提取特征的局部相关性从而提高特征提取的准确度.采集到的网络数据通过多层"卷积层-下采样层"的处理对网络中正常行为和异常行为的特征进行深度刻画,最后通过多层感知机进行正确分类.KDD 99数据集上的实验表明,文中提出的卷积神经网络模型与经典BP神经网络、SVM算法等相比,有效提高了入侵检测识别的分类准确性.As an effective algorithm of deep learning,deep convolutional nets have brought about breakthroughs in processing images,video,speech and audio,etc.In this paper,a convolutional neural network（CNN）modeling approach was used for intrusion detection,the convolution kernel was selected and convoluted with the data to extract local correlation,to improve the validity and efficiency of feature extraction.Through multiple ＂ convolution-downsampling＂ layer,collected feature was used to represent the normal and abnormal user behavior deeply.Finally,the multi-layer perception was used to classify these features.The experiment results on the KDD 99 data set show that,compared with the classical intrusion detection algorithms,the new model can improve the classification accuracy in the intrusion detection and recognition tasks.

关 键 词：入侵检测 深度学习 卷积神经网络 特征提取 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]