基于网络流量领域知识在深度学习中的模型构建方案

In order to improve the classification effect of network traffic model, this paper proposes a method that combines the domain feature of network traffic with deep learning. This model uses the domain features of network traffic, and also constructs a network layer auto-cross that automatically learns the cross features, and uses the attention mechanism to select important features to enrich low-order features and enhance the model\'s expression effect. Layers and auto-cross layers are used to handle domain features. Finally, convolutional neural networks are used to learn higher-order features to improve the generalization ability of the model. The two are fused to improve the model\'s ability to classify network traffic. Finally, experiments are performed on two data sets, and multi-class AUC is used as an evaluation index. The validity of the new model is verified through experimental comparison.