AN ENHANCED ENSEMBLE LEARNING FRAMEWORK FOR AUTOMATED ANDROID MALWARE DETECTION IN CYBERSECURITY

Abstract

Human life has changed from real-world to virtual worlds due to recent advancements in computer technology. Malware is superfluous software that is often used to initiate cyberattacks. Advanced packaging and obfuscation techniques are still being used by malware strains to evolve. These methods complicate the categorisation and detection of malware. To successfully battle emerging malware strains, new methods that vary from traditional systems should be used. All complicated and novel malware strains cannot be detected by machine learning (ML) techniques. The deep learning (DL) approach may be a viable way to identify every kind of malware. In this research, the Optimal Ensemble Learning Approach for Cybersecurity (AAMD-OELAC) approach for Automated Android Malware Detection is presented. The automatic categorisation and detection of Android malware is the main goal of the AAMD-OELAC approach. The AAMD-OELAC approach preprocesses data at the preliminary stage in order to do this. Three machine learning models—the Regularised Random Vector Functional Link Neural Network (RRVFLN), the Kernel Extreme Learning Machine (KELM), and the Least Square Support Vector Machine (LS-SVM)—are used in the AAMD-OELAC technique's ensemble learning process for Android malware detection. Lastly, the three DL models' optimum parameter tuning is achieved by using the hunter-prey optimisation (HPO) technique, which also contributes to better malware detection outcomes. A thorough experimental investigation is carried out to demonstrate the superiority of the AAMD-OELAC approach. The simulation results demonstrated the AAMD-OELAC technique's superiority over other methods already in use.