Android malware family classification with machine learning

As the popularity of Android mobile operating system grows, the number of software developed to harm the users of this system increases. Therefore, many studies have been done to detect malicious Android software. Apart from the classification of Android software as malicious or benign, classification of the malicious software into their families is also very important in terms of the security of the Android operating system. In this study, a machine learning based classification system is developed that analyzes malicious Android software and estimates the family of them. The developed system detects the requested permissions and API calls of the malicious Android software and uses them as features in machine learning algorithms to classify malwares. The performance of the system is investigated using various data sets and the evaluation results show that all classification algorithms classified the malware with a high accuracy. In addition to this work, a study of detecting an unknown malware which belongs to a family that had never seen before is made and these unknown malwares are classified with a high success rate.