Malware Family Detection in Android with machine learning-based methods

Android is an open-source and popular operating system that has attracted the attention of many malware writers. Nowadays, the speed of production and publication of Android malware has increased dramatically. For this reason, researchers are trying to come up with new methods to more accurately detect malware. Identifying the malware family will make us stronger against the threats of different types of malware and prevent future damage. Therefore, this paper presents a multiclass classification method to detect malware families. In the proposed method, preprocessing was performed to rank the features and select the effective features. Then, clustering was performed on the dataset (post-processing) to build a more accurate detection model. Finally, using the SVM algorithm, a model was presented for classifying malware families. Finally, the proposed method is capable of detecting malware families with an average accuracy of 96.56%.

Keywords:

Android Malware , Android Application Analysis , Family Detection