ABSTRACT In the rapid development of network technology, at the same time, a variety of network intrusion means are also emerging and network security, especially intrusion detection, is facing great challenges. Traditional intrusion detection algorithms identify intrusions using either traffic characterisation or machine learning. However, the problem is that these efficiencies do not meet current realities, especially when handling high‐dimensional anomalous data involving unexpected and unanticipated attacks. Therefore, to enhance efficiency and accuracy, we propose an intrusion detection method utilising a transformer and long short‐term memory (LSTM). We incorporate transformer and LSTM modules into intrusion detection models to identify network intrusions. Transformers and LSTM are able to obtain key data information, aiding in distinguishing abnormal data from normal data. In this paper, we conducted experiments on the publicly available CICIDS2017, UNSW‐NB15 dataset and validated them using commonly used evaluation metrics in the field of anomalous traffic detection. The experimental results demonstrate that our proposed model outperforms existing methods.
Dong et al. (Thu,) studied this question.