Comparative Analysis of Malware Android Apps Detection Using Machine Learning Approach

Abstract

Android is one of the most developed intelligent operating systems on mobile devices and has taken the
most part of the cell phone market. The rapid evolution of mobile devices technology has increased the number of
mobile malware in the application market, particularly when Android OS is widely adopted in the mobile devices.
These Android malicious applications hidden behind the benign applications pose a serious threat to the Android
platform. The end users and service providers are affected by malwares that are spreading around the world. In
order to mitigate the threats posed by a malware app, there is a need for developing applications that detect
Android malwares.
This paper work focuses on the identification of Android malware using machine learning approaches. The
objective of this paper is to classify the android application into benign or malware application. The proposed
system utilizes the features of collected random samples of benign and malware apps to train the classifiers. The
system extracts permissions used in the android applications as its features. With the extracted features, machine
learning approaches are used to classify the applications as benign or malicious. To classify the android
applications K-nearest neighbor, Decision trees, Support vector machines algorithms are used and the
performance of the classifier is calculated.