An efficient classification of malware behavior using deep neural network

Issue title: Special Section: Green and Human Information Technology

Guest editors: Seong Oun Hwang

Article type: Research Article

Authors: Hai, Quan Tran^a | Hwang, Seong Oun^{b; *}

Affiliations: [a] Department of Electronics and Computer Engineering, Hongik University, Sejong, Korea | [b] Department of Software and Communications Engineering, Hongik University, Sejong, Korea

Correspondence: [*] Corresponding author. Seong Oun Hwang, Department of Software and Communications Engineering, Hongik University, Sejong, Korea. E-mail: sohwang@hongik.ac.kr.

Abstract: Malware detection have long become a challenge in research. The existing methods rely on malware signature which are proved not to be effective nowadays. The recent researches focus on using probabilistic model such as machine learning to detect the existence of malware. They, however, do not achieve such a good performance. Particularly, machine learning techniques still have an issue of high feature engineering overhead. In this paper, we propose a deep learning method to detect malware based on their malicious behavior. Through experimentation, we show that our method can achieve a very high accuracy rate of 98.75 in F1 measure, compared to state of the art methods.

Keywords: Malware classification, deep neural network, security

DOI: 10.3233/JIFS-169823

Journal: Journal of Intelligent & Fuzzy Systems, vol. 35, no. 6, pp. 5801-5814, 2018