Bidirectional LSTM Models for DGA Classification

Giuseppe Attardi, Daniele Sartiano

Abstract: The paper describes our submission to the shared task on DGA classification at DMD 2018. The approach is based on a Deep Learning architecture using bidirectional LSTM neural networks. Similar models are used in both the tasks, the first one is to identify the DGA generated domain name and the second one is to detect and categorize the DGA generated domain name to their botnet family.