ANDROID MALWARE DETECTION USING MACHINE LEARNING

Abstract

The rapid growth in Android smartphone usage has led to a corresponding increase in Android malware attacks, posing significant threats to user privacy, data security, and device integrity. Traditional signature-based malware detection techniques are effective against known threats but often fail to detect new, modified, or polymorphic malware. This research investigates the application of machine learning techniques for Android malware detection by analyzing features extracted from both benign and malicious applications. The study evaluates the performance of several classification algorithms, including Naïve Bayes, Logistic Functions, Hoeffding Tree, J48, and LMT, using Android malware datasets such as Drebin 215. Feature selection methods are employed to improve detection accuracy while reducing false positives. The findings indicate that machine learning provides a more effective and adaptive approach to malware detection than conventional methods, with the enhanced Naïve Bayes classifier achieving an improvement in detection accuracy from 85% to 92%. The proposed approach demonstrates the potential of intelligent malware detection systems to enhance Android device security by accurately identifying malicious applications and adapting to evolving cyber threats.