Using Deep Learning to Detect Malicious URLs
Yuchen Liang, Xiaodan Yan · 2019 IEEE International Conference on Energy Internet (ICEI) · 2019
With the development of energy internet, cyberattacks pose a great threat to the distribution of electricity. Specifically, malicious domains can cause tremendous damages to network security. Many researchers have tried to tackle the task of detecting DGA-generated domains with different approaches. In this paper, an algorithm based on the Deep Bidirectional LSTM model is constructed to address this problem. Before the final experiment with the deep learning model, this paper also presents traditional machine learning methods that classify malicious domains based on lexical features for comparison. The result shows that the DBLSTM classifier can achieve a 98.6% accuracy, performing much better than conventional machine learning approaches.