Network Abnormal Behavior Detection Based On Deep Learning

With the rapid development of information technology,the unprecedented prosperity of network technology,on the one hand,to promote the continuous development of human civilization,and on the other hand,it breeds all kinds of network security problem.In the context of vast data,the traditional network security measures are not enough to cope with the increasingly complex network environment.The reason for this is that it can't withstand attacks from within the system,nor can it be monitored in real time for attacks.Therefore,how to conduct real-time and comprehensive detection of vast quantities of data is an important research trend.In recent years,this problem has been solved by the detection techniques of network abnormal behavior,which is based on dynamic defense.At the same time,various network algorithms based on deep learning show the obvious advantages of large data processing.Therefore,based on the comprehensive analysis of the detection and deep learning techniques of network anomaly,this paper puts forward the model of detecting network anomalies based on deep structure.First of all,based on convolutional neural network(CNN)is used to the depth of the learning algorithm of network anomaly behavior training set features of training data,then use support vector machine(SVM)classification algorithm sets of test data.The convolution neural network of this paper adopts the modified lenet-5 network structure,and the support vector machine adopts two classifications and uses the particle swarm algorithm to tune it.In this paper,a number of experiments are carried out for the above algorithm model in the experimental part.In this paper,the experimental results show that the algorithm is feasible and accurate.Based on the above algorithm model,this paper designs and implements a cloud forecasting system based on multi-terminal CNN-SVM feedback supervised model.Inthis paper,the system architecture,system flow,module division and the concrete realization of each module are described in detail.At the same time,this paper introduced the system in the experimental part of the system function test,the result of the experiment proves that this system can carry on the effective detection to the network behavior and positioning.