Permissions-based Android malware detection using machine learning

Android applications are evolving quickly, and both consumers and developers are growing in number. This popularity increases Android system vulnerabilities and attacks. Researchers have proposed various approaches to detect Android malware. One of these approaches is to utilize machine learning to detect Android malware based on permissions. This study developed a permissions-based Android malware detection model using machine learning. It aims to identify the significant permissions list that can be used to distinguish between benign and malware apps. A dataset containing permissions for both types of apps was collected to compare feature selection algorithms (IG, RFE, CHI, GA) with machine learning algorithms (SVM, DS, RF, GB, XG) to identify the best prediction model. The results showed that the SVM model achieved the highest accuracy of 98.88% using RFE with 13 permissions. The model was able to complete the detection process in 12.00 milliseconds.