De-LADY: Deep learning based Android
malware detection using Dynamic features 1Sardar
Patel University of Police, Security
and Criminal Justice, Jodhpur, India 2National
Institute of Technology, Raipur,
India 3School
of Computing Science and Engineering, VIT Bhopal University, Bhopal, Madhya Pradesh,
India 4Korea
Communication Agency, South
Korea seiilson@kca.kr Abstract Popularity and market share of Android operating system has given significant rise to malicious apps targeting it. Traditional malware detection methods are obsolete as current malwares are equipped with state of the art obfuscation methods to hide their intent from scanning engines. In this paper, we propose De-LADY (Deep Learning based Android malware detection using DYnamic features) an obfuscation resilient approach. It utilizes behavioral characteristics from dynamic analysis of an application executed in emulated environment. The proposed approach is evaluated against 13533 applications from categories such as banking, gaming and utilities. De-LADY is effective with 98.08% detection rate and 98.84% F-measure. Furthermore, it outperformed existing machine learning approaches. Keywords: Android, Malware detection, Code
obfuscation, Familial classification +: Corresponding author: Vikas Sihag Journal of Internet
Services and Information Security (JISIS), 11(2):
34-45, May 2021 Received:
February 5, 2021; Accepted: May 2, 2021; Published: May 31, 2021 DOI:
10.22667/JISIS.2021.05.31.034 [pdf] |