Android Malware Detection using Machine Learn...

Android Malware Detection using Machine Learning

Author Name : Ankashu Deepika, S.Sreehaas, Yerupula V. Pahvvan , Ms. G. Mounica

ABSTRACT

This article addresses the problem of increasing malicious Android applications that pose a threat to users' privacy and control over their devices. Despite efforts to protect against these apps, they continue to circulate and cause harm. The article proposes a solution using malware analysis and reverse-engineering of Android applications, employing both static and dynamic analysis techniques.The objective of the article is to protect users from malware by detecting it through machine learning algorithms. By reverse engineering the apps, the code is analyzed to identify any malicious activities, such as unauthorized data access or control of the device. Permissions within the code are extracted to create a feature dataset, where the presence or absence of proper permissions determines whether an app is classified as malware or goodware.Features like permissions and app components are extracted and represented as a feature vector with class labels in CSV format. To streamline the feature selection process, a genetic algorithm is applied to identify the most relevant features. The resulting optimized feature set is then used to train SVM and NN classifiers. Static features are obtained from the AndroidManifest.xml file, which contains crucial information about the apps, and the Androguard tool is employed for disassembling the APKs and extracting static features.The advantage of the article includes improved detection accuracy through a novel feature selection algorithm and the ability to identify new variants of Android malware using machine learning and static and dynamic analysis. The utilization of genetic algorithms optimizes the detection process, reducing execution time.

KEYWORDS: Android Malware Detection; Machine Learning, Python;Static analysis; API-calls; Permissions; Gated Recurrent Unit.