Malicious Traffic Classification Model Based on Hypergraph-LSTM Neural Network

With the development of the Internet and the increasing degree of people's dependence on the network, the problem of malicious traffic flooding has gradually appears. How to effectively identify malicious traffic has become an important challenge in the field of network security. In malicious traffic classification research, deep learning can effectively represent the characteristics of traffic data, to improve the classification ability of traffic data. However, the existing malicious traffic classification methods cannot fully extract data features and have certain limitations. In this paper, a malicious traffic classification model based on hypergraph structure and LSTM network, Hypergraph-LSTM neural network, is constructed. In the model, the traffic data is constructed as a hypergraph, enabling a richer representation of information and the implicit integration of more similar data within the model. The LSTM module is utilized to extract temporal features from data. The design effectively combines representations containing both spatial and temporal semantics. Tests on several publicly available datasets show that Hypergraph-LSTM outperforms other current methods.