Research Of Intrusion Detection Technology Based On Deep Learning

With the rapid development of computer and network information technology,the Internet has become an important way for people to obtain outside information.At the same time,the development of network technology has also brought some network information security issues,making security monitoring more and more important.At present,the network environment is becoming more and more complicated,and the data of network intrusion behavior has characteristics such as high feature dimensions,large amount of data,and large redundancy.Traditional intrusion detection models rely heavily on feature selection,which makes them very limited in processing high-dimensional,non-linear mass data.Deep learning technology can effectively extract the deep features of intrusion data from massive data,so that it can better detect intrusion behavior.This paper adopts deep belief network as the feature learning method for massive data,proposes an improved deep learning model applied in the field of intrusion detection,and designs an improved intrusion detection model.The main research work of the paper is as follows:1.In order to solve the problem of complex data types and feature redundancy in the field of intrusion detection,the paper uses deep belief network to learn and reduce dimensionality of data features.The paper proposes an adaptive learning rate method to improve the convergence speed and accuracy of network training,and designs a improved deep belief network model(IDBN)based on adaptive learning rate.The paper determines the network structure of the feature learning model for intrusion detection(122-90-60-40-20).Experiments show that the detection accuracy of the IDBN model is 2.19% and 7.97% higher than the auto-encoder network and principal component analysis-20 models,respectively.2.The initial connection weights of the IDBN model are directional and prone to premature convergence,so the paper proposes an intrusion detection algorithm based on IDBN.This algorithm uses a genetic algorithm-particle swarm optimization(GA-PSO)hybrid optimization algorithm that introduces the crossover and mutation operations of genetic algorithm(GA)based on the research of particle swarm optimization(PSO)algorithm to optimize the initial network parameters of the IDBN model.Experiments show that the detection accuracy of IDBN model using GA-PSO hybrid optimization algorithm is 0.53% and 1.39% higher than GA and PSO algorithms,respectively,and its false alarm rate is 2.26% and 0.46% lower than GA and PSO algorithm,respectively.3.The paper designs and implements a deep learning-based intrusion detection model(GA-PSO-IDBN-IS),which mainly includes three parts: data preprocessing,GA-PSO-IDBN feature learning,and an improved softmax classifier.Compared with the three intrusion detection models such as convolutional neural network,long short term memory recurrent neural network and GA-PSO-IDBN-softmax,through the two classification and multi-class detection experiments,it shows that the proposed intrusion detection model has higher detection accuracy and lower false alarm rate.