Detecting android malware and Prevention Using Supervised Learning

Abstract

The Android smartphone's growth may be attributed to the phone's open-source design and high performance. Malware has been created partially because of Android's widespread use. When it comes to smartphones, Android is the most popular OS. That's why there's so much malicious software aimed at this system. Malicious software may be identified as such by analyzing its permission attributes. But this is a complex issue to solve. In this research, we use a golden jackal optimized support vector machine (GJOSVM) to classify software and evaluate whether or not it presents a threat. To achieve this goal, a dataset including 2850 sections of malicious software and 2866 sections of benign software was generated. Each piece of software in the dataset has 112 permission characteristics, and there is also a class feature that indicates whether or not the program is harmful. Each phase of the training and testing procedures used 10-fold cross-validation. The effectiveness of the models was measured using accuracy, F-1 Score, precision, and recall.