Title: Fast classification of malicious traffic in networks based on multi-stage neural network

Authors: Zhenyu Yang; Ranran Yin

Addresses: Department of Urban Rail Transit and Information Engineering, Anhui Communications Vocational & Technical College, Hefei, Anhui, China ' Department of Urban Rail Transit and Information Engineering, Anhui Communications Vocational & Technical College, Hefei, Anhui, China

Abstract: In order to overcome the problems of slow convergence speed, low false alarm rate and high cross-entropy loss in traditional fast classification methods for malicious network traffic, a new fast classification method of malicious traffic in networks based on multi-stage neural network is proposed. Firstly, collect network traffic data from multiple dimensions and normalise and deduplicate it. Then, a multi-stage neural network is composed of hypergraph neural network and recurrent neural network, which is used to extract spatial features and time-dependent features. Finally, the features are fused and input into the softmax function for classification, and the cross-entropy loss function is used to optimise the classification results. Through experiments, it has been proven that the method proposed in this paper has a fast convergence speed, low false alarm rate and low cross-entropy loss in classification, indicating that the traffic classification accuracy of this method is high.

Keywords: multi-stage neural network; malicious network traffic; fast classification; hypergraph neural network; recurrent neural network; spatial features; time-dependent features.

DOI: 10.1504/IJIPT.2024.146275

International Journal of Internet Protocol Technology, 2024 Vol.17 No.4, pp.191 - 200

Received: 12 Feb 2025
Accepted: 30 Mar 2025
Published online: 14 May 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article