Research On Intrusion Detection Model Based On ResNet

With the rapid development of mobile internet,the security situation of network space is becoming more and more serious.How to effectively monitor and protect network intrusion has become the research focus in the field of network security.With the arrival of the era of big data and the diversification of network attack means,traditional intrusion detection methods can no longer satisfy the requirements of today's network environment.As a new frontier technology,Deep Learning has developed rapidly since its emergence,and has been successfully applied in various academic fields.It provides a new idea for the development and change of intrusion detection technology.Since machine learning was first proposed,it has been widely used in various industry sectors.It also injects new blood and power into intrusion detection technology.It changes the mechanism of intrusion detection from passive to active.Deep Learning is a new technology after Machine learning has developed to a certain stage.It has a superior performance in the field of image processing.This paper combines the idea of Deep learning with intrusion detection technology,and proposes an intrusion detection model based on ResNet.The main work of this paper is as follows: Firstly,UNSW-NB15 is selected as the dataset of intrusion detection model,which is analyzed and compared with the widely used intrusion detection benchmark dataset,and then discussed the advantages of UNSW-NB15 dataset.Secondly,the Standardized Processing of dataset,such as deleting redundant attribute features,numeralization of attribute features of symbol type,normalization of the whole data,and so on.Thirdly,this paper uses the method of transforming network traffic data into images,and expresses one-dimensional network traffic data in the form of two-dimensional gray image,and uses image recognition to solve intrusion detection problems,so as to apply Deep learning technology to intrusion detection.Fourthly,the convolutional neural network and its classical models(LeNet-5,AlexNet and ResNet)are deeply analyzed,and the intrusion detection model based on ResNet is constructed.Fifthly,through simulation test,the performance of intrusion detection model based on LeNet-5,AlexNet and ResNet is analyzed and compared;The performance of IDS based on ResNet is compared and analyzed under different activation functions and network parameters.,and then adjust the structure and parameters of the model to improve the overall performance of the model.