Intelligent Android Malware Detection System Using Ensemble Machine Learning

Abstract

The privacy and security of users are seriously threatened by the exponential growth of Android devices and the rising prevalence of malware. Because malware is always developing, conventional signature-based malware detection technologies are becoming ineffective. As a result, the demand for sophisticated malware detection systems that can accurately identify fresh and undiscovered malware strains is rising. In order to improve detection accuracy and robustness, this study suggests an intelligent Android malware detection system that makes use of ensemble machine learning techniques. The suggested approach uses an ensemble model created by combining different machine learning algorithms, such as gradient boosting, random forests, support vector machines (SVM), and decision trees. Each base model is trained using a wide variety of features that are taken directly from Android applications, including permissions, API requests, and manifest data. In order to reach a final determination regarding whether an application is malicious or benign, the ensemble model combines the predictions from individual models. An extensive dataset made up of both known and undiscovered malware samples is utilised to assess the system's performance. The experimental findings show that in terms of accuracy, precision, recall, and F1-score, the ensemble model surpasses individual machine learning techniques. The ensemble model successfully lowers false positives and false negatives while achieving a high detection rate for both known and undiscovered malware. The suggested approach also demonstrates remarkable generalisation skills, enabling it to adjust to fresh and undiscovered malware types.