MALICIOUS NETWORK TRAFFIC DETECTION METHOD BASED ON BIDIRECTIONAL GRU AND CNN

To solve the problems of insufficient accuracy and poor generalization of current malicious network traffic detection technology, a malicious network traffic detection method based on bidirectional GRU and CNN is proposed. Bidirectional GRU and CNN were used to extract temporal and spatial features of network traffic data in parallel, and self-attention mechanism was added to calculate the importance of features. Experiments were carried out on CIC-IDS2017 dataset. The results show that the accuracy of the detection method in multi-class classification and binary classification are 99.77% and 99.82% respectively, which is superior to other detection methods.