ANALYSIS AND DETECTION OF MALWARE IN ANDROID APPLICATIONS USING MACHINE LEARNING

Abstract

The rapid expansion of Android applications has significantly increased the risk of mobile malware, necessitating advanced and efficient detection mechanisms. This study presents a hybrid machine learning-based framework for Android malware detection that integrates both static and dynamic analysis to improve classification accuracy and reliability. Static features such as permissions and intent filters are extracted through reverse engineering of application files, while dynamic features, including API call sequences, are captured during runtime execution in a controlled environment. Unlike conventional approaches that rely on single-feature analysis, the proposed system combines these heterogeneous features to capture both structural and behavioral characteristics of applications. To enhance detection performance, the extracted features are preprocessed and transformed into structured representations suitable for machine learning models. A Support Vector Machine (SVM) classifier is employed as the primary detection model due to its robustness in handling high-dimensional data and classification boundaries. Experimental observations indicate that dynamic features provide better detection capability compared to static features, while the hybrid approach achieves superior overall performance. The proposed method demonstrates improved accuracy, reduced false positives, and better generalization, making it suitable for realtime Android malware detection and mobile security applications.