AN ENHANCED HYBRID ARCHITECTURE FOR DETECTING MALWARE IN ANDROID APPS USING MACHINE LEARNING AND DEEP LEARNING TECHNIQUES

Smartphones are multipurpose gadgets with a wide range of capabilities and constant communication. However, their adaptability, also leaves them open to viruses, endangering user security, privacy, and their financial security. The identification and classification of Android malware has consequently emerged as a critical area of cybersecurity research interest. A rise in malicious assaults on the Android platform has been caused by the high demand for the Android operating system, which has gained malware developers' attention. By stealing sensitive data and degrading the efficiency of devices, such attacks may seriously harm the user. This research paper examines the challenges of identifying Android malware. This study aims to identify malicious and benign files from large datasets using machine learning (ML) and deep learning (DL) techniques to develop efficient, accurate, and robust models for malware detection. We propose a novel ADAX-NETBoost approach, that outperforms existing classification methods with an impressive detection accuracy of 99.34% and 99.21% on Android Malgenome and Drebin dataset, respectively. The experimental results validate the effectiveness of our proposed approach in accurately detecting Android malware, outperforming earlier studies. Keywords: Hybrid Ensemble Learning Models, Classification Model, Permissions, API Calls, Android Malware.