Malicious Domain Name Detection Model Based on CNN-GRU-Attention
Yanshu Jiang, Mingqi Jia, Biao Zhang, Liwei Deng · 2021
Domain Generation Algorithm (DGA) domain name detection is one of the key technologies for detecting botnet C&C communications. It is well known that malicious websites can cause great harm, and from individuals to countries will be affected to varying degrees. Aiming at the problems of low detection accuracy and high complexity of traditional detection methods, this paper proposes a malicious domain name detection model (CNN-GRU-Attention). The model first used the CNN neural network to extract the spatial features of the domain name data; then used the GRU neural network to extract the temporal features of the domain name data; finally used the attention mechanism to improve the detection accuracy of the domain name. In the experiment, this article used Bigrams, LSTM artificial neural network, GRU neural network, LSTM-GRU four models to compare with the CNN-GRU-Attention model. The experimental results showed that the CNN-GRU-Attention model had better convergence and higher accuracy.