Android Malware Detection Using Machine Learning

Now a days, malware has become a more and more concerning matter in the security of information and technology proven by the huge increase in the number of attacks seen over the past few years on all kinds of computers, the internet and mobile devices. Detection of zero-day malware has become a main motivation for security researchers. Since one of the most widely used mobile operating systems is Google's Android, attackers have shifted their focus on developing malware that specifically targets Android. Many security researchers used multiple Machine Learning algorithms to detect these new Android and other malwares. In this paper, we propose a new system using machine learning classifiers to detect Android malware, following a mechanism to classify each APK application as a malicious or a legitimate application. The system employs a feature set of 27 features from a newly released dataset (CICMalDroid2020) containing 18,998 instances of APKs to achieve the best detection accuracy. Our results show that the methodology using Random Forest has achieved the best accuracy of 98.6% compared to other ML classifiers.