Research on Deep Learning Models Combining Transformer and BERT: Application in Network Intrusion Detection
Zhongqiu Fang, Ziwei Hong · 2024
In response to the low recognition accuracy, complex preprocessing techniques, and unstable detection accuracy in network intrusion detection, this paper proposes a deep learning model based on the combination of Transformer and BERT. Firstly, based on the basic theories of Transformer and BERT, an intrusion detection model was established to obtain detailed information on network traffic characteristics, intrusion behavior types, and detection performance. Secondly, in order to eliminate the interference of these issues, a principal component analysis strategy was proposed. Furthermore, based on the characteristics of Transformer neural networks, an intrusion detection model was constructed that comprehensively considers the influence of network traffic characteristics and intrusion behavior types, providing a more accurate description of network intrusion behavior. Finally, an intrusion detection system based on Transformer and BERT was built, and the effectiveness of the proposed algorithm was verified through numerical examples.