Research on Traffic Classification and Identification of Network Abnormalities based on Deep Learning Algorithms

In an increasingly complex network environment, accurately and effectively identifying abnormal network traffic has become a key challenge in ensuring network security. Traditional manual detection methods are time-consuming and prone to errors, therefore there is a need for a Automated solutions. This article proposes a hierarchical classification based on deep learning Detect abnormal network traffic. In particular, the convolutional neural network architectures Inception V3 and Densenet121 are used to analyze network traffic across multiple dimensions, including traffic speed, traffic size, compactness of anomalous patterns, and spatial distribution. Experiments have shown that Densenet121 outperforms Inception V3 in terms of flow velocity and spatial distribution, while Inception V3 outperforms Inception V3 in terms of classifying traffic size and compactness. The accuracy, sensitivity, and specificity of the estimates demonstrate the effectiveness of the proposed model. This study is a comparative analysis based on CNN architecture, providing valuable reference for intelligent network anomaly detection.