MACHINE LEARNING-BASED FRAMEWORK FOR AUTOMATIC MALWARE DETECTION USING ANDROID TRAFFIC DATA

One of the greatest challenges facing various organizations and institutions is information security. Attackers have devised means to steals mobile user identity by developing malware that might be inadvertently installed by users due to the open source nature of android operating system causing financial loses. Although various machine learning algorithms have been proposed recently for malware detection, it is challenging to detection malicious apps with single classification model. In this paper, we propose to detect malicious apps in android traffic using four (4) different machine learning algorithms. The proposed approach was evaluated on comprehensive and publicly available dataset. The result obtained shows that decision tree and tree based ensemble algorithms produced superior results when compared with support vector machine and logistic regression models. The results suggest the impact of multiple classification algorithms to improve the performance of malware detection system. The finding can be utilized to guide security expert on the use of machine learning methods to detect malicious software