Research And Application Of Generated Anti-Neural Network In Intrusion Detection

With the development of the network and the leap-forward progress of information and communication technology,especially the large-scale commercial use of 5G communication technology,network security issues are facing major challenges.As an active network security protection technology,intrusion detection technology can provide network systems with more real-time,efficient,and secure protection,and strive to respond to and intercept network systems before they are damaged.Therefore,intrusion detection is an indispensable part of network security protection.The intrusion detection system is usually constructed by using machine learning methods to train the classification model.However,the training sample data of intrusion detection technology is currently updated slowly,and the confidentiality of the data is inconvenient for disclosure.This has a large impact on the accuracy of the model.A careful study of the current domestic and foreign for quantitative association rules mining research status and related mining methods was made the research status of intrusion detection systems and generative adversarial networks at home and abroad,as well as related application methods.Based on this,the research work of combining generative adversarial networks with intrusion detection is carried out.According to the powerful data generation capability of generative adversarial neural networks,a method for generating generative adversarial networks for intrusion detection data generation is proposed,and the data generated by this method is compared with the original data.This method mainly has the following main innovations:First,for the first time,the generative adversarial neural network commonly used in computer vision is used to generate intrusion detection data sets.Second,provide solutions to the problem of low data volume due to difficulties in network stream data collection,which in turn leads to low accuracy of the training model.Thirdly,the small sample data set can be used to generate data through the generation network to train an effective classification model.In order to verify the validity of the generated data,the same discriminative network is trained using the generated data and the real data,and the generated data and real sample data are mixed according to a certain ratio.The experimental results show that the generated data can basically match the training results of the real samples on the training model,And the classification results on the small sample data are better than the classification effect of the real data training model.