Support Vector Machines Based Android Malware Detection Using Permissions, API Calls, and Library Lists

In the context of increasing malware development in Android operating systems. The use of machine learning in malware detection and prevention has resulted in breakthroughs and efficiency in identifying complex patterns and even new patterns that have not appeared before. This article focuses on the problem of detecting two malicious (malware) and benign (benign) labels and presents a method to classify Android malware based on permissions, function calls, and library lists. The goal is to build a classification model that can distinguish between files, applications, or behaviors that are malware and benign. The data were collected from the Android application file (APK) and converted into a suitable form for classification. The recommendation system includes the following components: features extraction, feature selection, malware detection based on deep learning, and prediction. We evaluated the performance of the proposed system by comparing it with some of the most advanced detection methods. The test results show that the system achieves a significant improvement, and the F1-score ratio for the SVM model is 0.97028, demonstrating the feasibility of the system in practice.