FLANNdroid: A framework for malware detection from Android devices.

Abstract

On the daily basis, Android based devices has gain popularity in the life of users. Everyone depend upon the smartphone apps for its daily work. 3.3 million apps present in the official store of Android. Android apps depend upon the permission-model for its proper working. In this research paper, 50,000 Android apps are collected from real-world. Dynamic analysis is performed on these collected apps and extract features from them. Feature selection techniques are applied on extracted features and select relevant features from them. Further, hybrid machine learning algorithm is proposed which is the based on the principle of functional link artificial neural network with genetic algorithms. Empirical result reveals that proposed algorithm is achieved an detection rate of 98.7% while testing is performed on real-wrold apps.