LSTM Network Based Malicious Domain Name Detection
Gurpreet Singh Josan1, Jagroop Kaur2
1Gurpreet Singh Josan*, Department of Computer Science, Punjabi University, Patiala, India.
2Jagroop Kaur, Department of Computer Science and Engineering, Punjabi University, Patiala, India.
Manuscript received on July 20, 2019. | Revised Manuscript received on August 10, 2019. | Manuscript published on August 30, 2019. | PP: 3187-3191 | Volume-8 Issue-6, August 2019. | Retrieval Number: F8809088619/2019©BEIESP | DOI: 10.35940/ijeat.F8809.088619
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Detecting malicious domain names attract lot of research in recent years. Researchers tried various text based, network traffic based and combination of these methods to detect malicious names. In this paper, we analyze the possibility of detection malicious names using deep neural network based models. Bidirectional LSTM network has been developed and trained on the dataset. Two tasks were experimented. First task was to identify malicious domain name and second task was to identify the class of domain name. Proposed method is able to perform well on task 1 producing 98.9% accuracy whereas on task 2 it is able to achieve accuracy of 69.7% only.
Keywords: Botnet, Deep Neural Network, Malicious domain name classification.