AUTOMATED ANDROID MALWARE DETECTION USING OPTIMAL ENSEMBLE LEARNING APPROACH FOR CYBER SECURITY

Abstract

Current technological advancement in computer systems has transformed the lives of humans from real to virtual
environments. Malware is unnecessary software that is often utilized to launch cyberattacks. Malware variants are still
evolving by using advanced packing and obfuscation methods. These approaches make malware classification and
detection more challenging. New techniques that are different from conventional systems should be utilized for
effectively combating new malware variants. Machine learning (ML) methods are ineffective in identifying all
complex and new malware variants. The deep learning (DL) method can be a promising solution to detect all malware
variants. This paper presents an Automated Android Malware Detection using Optimal Ensemble Learning Approach
for Cybersecurity (AAMD-OELAC) technique. The major aim of the AAMD-OELAC technique lies in the automated
classification and identification of Android malware. To achieve this, the AAMD-OELAC technique performs data
preprocessing at the preliminary stage. For the Android malware detection process, the AAMD-OELAC technique
follows an ensemble learning process using three ML models, namely Least Square Support Vector Machine (LSSVM),
kernel extreme learning machine (KELM), and Regularized random vector functional link neural network
(RRVFLN). Finally, the hunter-prey optimization (HPO) approach is exploited for the optimal parameter tuning of
the three DL models, and it helps accomplish improved malware detection results. To denote the supremacy of the
AAMD-OELAC method, a comprehensive experimental analysis is conducted. The simulation results portrayed the
supremacy of the AAMD-OELAC technique over other existing approaches.