Malicious HTTPS Traffic Classification Algorithm Based on DCGAN_1D-CNN
Wen Luo, Zhaohui Liu, Rui Qi Zhao, Jianhua Chen, Xin Deng · 2021 IEEE Conference on Telecommunications, Optics and Computer Science (TOCS) · 2021
Aiming at the problems of fast classification and unbalanced data classification of encrypted malicious traffic in the internet, a classification method of encrypted malicious based on DCGAN_1D-CNN model is proposed.DCGAN-IDCNN uses the idea of generating confrontation to generate a few samples in the data level to improve the lack of original training samples and sample imbalances,and use the ID-CNN to train the original sample and the generated few samples,so as to accurately classify the encrypted malicious.The experimental results on the public dataset CICAndMal2017 show that the F1 value reaches 96.55% in the binary classification experiment, which is at most higher 11.77% than that of using 1D-CNN model alone,indicating that the DCGAN_1D-CNN model has stronger ability to classify encrypted malicious traffic.