Character-level domain name generation algorithm based on ED-GAN

In response to the need to obtain the latest counterfeit domain name samples in time to prevent new counterfeit domain name attacks in high-level security requirement scenarios, this chapter proposes a character-level domain name generation method based on ED-GAN. By combining Encoder of domain names, Decoder of domain name vectors, and Generative Adversarial Network (GAN), an ED-GAN-based domain name generation algorithm is proposed to generate realistic data by learning from real counterfeit domain names and using the generated data for training of detection models to achieve detection and prediction of emerging and potential counterfeit domain names. The generated data is used to train the detection model to detect and predict emerging and potential counterfeit domain names. Finally, the feasibility of the ED-GAN-based character-level domain name generation algorithm and the effectiveness of the generated data are demonstrated through experimental performance comparisons with various classifiers.