Mobile Malware Detection using Anomaly Based Machine Learning Classifier Techniques

Abstract

Smartphones have become essential in our daily lives. Many works can be done by using it like, browse the internet, and download many applications for each device through the available store. As a result, the number of malware applications downloaded also increases. These malware carries out various activities behind the scenes, such as breach of confidentiality, breach of privacy, loss of confidentiality, system breakdown, theft of sensitive information, etc. Many types of research and studies have proposed different techniques to detect malicious programs, but these measures contain weak points, which are illustrated by efficiency, speed, and lack of comprehensiveness. In this paper, a proposed system is designed and implemented to detect malware in smartphones using anomaly detection technology that begins to extract the important features that play an effective role in detecting malicious code and applying machine learning algorithms. The proposed system has been tested using a hybrid Genetic algorithm, and the Support Vector Machine data has been registered with an accuracy of (0.9282%). The experimental results indicate that the proposed system has a high average accuracy rate compared with other existing methods where there is a (0.8848%) average accuracy using Probabilistic Neural Network, while the average accuracies of (0.8835%) and (0.8715%) respectively with Support Vector Machine and K-Nearest Neighbors.