Research On Network Intrusion Detection Based On Deep Learning Model

With the development of the Internet,network security issues have emerged in an endless stream,and network attack methods have been constantly innovating,and higher requirements have been placed on network intrusion detection technology.The amount of network data continues to increase,and data characteristics are gradually changing.Traditional machine learning-based intrusion detection models have become increasingly difficult to meet current needs in the security protection of network environments.The deep learning method is a very popular method at present,it is widely used in image processing and natural language processing and other fields,and has achieved good results.At present,one of the main reasons for the high false alarm rate and low detection rate is that the characteristics of the data cannot be fully learned.In order to overcome this shortcoming,this article applies the deep learning model to the research of network intrusion detection.This paper proposes a model DNN-FM that can interactively learn low-level features and high-level features.The factorization machine is used to extract and learn low-level features,and the deep neural network(DNN)is used to extract and learn highlevel features,so this model can learn low-level features and high-level features interactively.After experiments on the KDD CUP99 data set,it is proved that the DNNFM model has a higher detection rate than the existing network intrusion detection model.In order to further improve the accuracy of classification,in view of the problem that the existing network intrusion detection model cannot fully learn the characteristics of network intrusion data,the model was redesigned,on the basis of the DNN model,an dilated convolution model and a GRU model are added,in which dilated convolution is used to increase the receptive field of information,high-level features are extracted from it,and the long-term dependence relationship between retained features is extracted using the gated recurrent unit(GRU)model,and then DNN is used for data features Fully study.Compared with classic machine learning classifiers,this model has a high detection rate.Experiments on the KDD CUP99,NSL-KDD and UNSW-NB15 datasets show that the model has leading performance.The experimental results are: the accuracy rate of using the KDD CUP99 data set on the DNN-FM model is 98.96%;In the redesigned model,the accuracy rate of using the KDD CUP99 data set is 99.78%,the accuracy rate of using the NSL-KDD data set is 99.53%,and the accuracy rate of using the UNSW-NB15 data set is 93.12%.The experimental results show that the model proposed in this paper has a high detection rate,which has certain reference significance for network intrusion detection research.