Research On Network Traffic Anomaly Detection Based On Convolutional Neural Network And Key Features

With the advent of the digital age,everything is interconnected,and information technology represented by artificial intelligence and big data is becoming more and more closely integrated with traditional industries.At the same time,the ever-increasing number of smart terminals generates a large amount of traffic data,making the network environment complex and diverse.This brings new challenges to network security protection,and traditional anomaly detection methods are difficult to adapt to the current network situation;At the same time,with the advancement of digital construction,network security is no longer an auxiliary role,but has become the foundation of digital construction.Therefore,it is necessary to study a new type of network traffic anomaly detection method.There are two main problems facing current network traffic anomaly detection: detection accuracy and efficiency need to be further improved;the identification of high-dimensional and complex traffic data needs further research.First,in order to improve the accuracy and efficiency of traffic anomaly detection,it is proposed to use an improved bilinear convolutional neural network for malicious traffic classification.This method makes full use of the powerful feature extraction ability and selflearning ability of the convolutional neural network to improve the detection accuracy and detection efficiency of abnormal traffic recognition.Through the construction of neural network structure,detection model pre-training and classification detection experiments,it is concluded that the abnormal traffic detection accuracy rate of this method can reach 97.8%,which has certain advantages compared with other detection algorithms.Second,in the face of high-dimensional network traffic data,in order to improve the efficiency of anomaly detection and analyze the importance of traffic security features,it is proposed to use Light GBM algorithm for network traffic feature selection and analysis.Try to use fewer features for flow detection to improve the detection efficiency of high-dimensional flow data.Through data preprocessing,key feature selection experiments and comparative experiments,it is concluded that using fewer traffic features,the detection efficiency is increased by 25%,and the detection accuracy is higher.This research work also has certain practical applications value.