Dynamic Android Malware Detection Using Light Gradient Boosting Machine

Growing popularity of Android smartphones, its vast user base, open architecture, vague permission management system, lack of security standards, absence of reliable app inspection, and increasing use of smartphones in sensitive areas like banking, security, identity management have made Android a lucrative target for malware developers. Android malware is becoming more and more sophisticated with the use of obfuscation techniques rendering the traditional signature based mechanism ineffective. A reliable, robust, efficient, and effective Android malware detection method is required to tackle with this emerging issue. Current techniques lack the are time inefficient and less reliable in terms of accuracy. In this work, we present a efficient, reliable, and robust Android malware detection technique based on light gradient boosting machine. We trained and tested the framework on dynamically extracted features. We have got the accuracy of 98.008%, recall of 99.194%, precision of 98.385%, and F1 score of 98.787%. On implementing the existing state of art techniques, we found that the proposed scheme outperforms them thereby establishing supremacy of our work.