A Comprehensive Analysis and Evaluation of Android Malware Prediction Using AI
Noor E Fatima, Hasan Faraz Khan · 2024
The increasing prevalence of Android malware necessitates the development of robust and accurate prediction models to safeguard users and their sensitive data. This study aims to develop reliable Android malware prediction models using a comprehensive dataset with multiple app permissions. Employing state-of-the-art ensemble learning techniques, we achieved a satisfactory accuracy of 81.47%. Notably, this surpasses all previous works, including deep learning approaches, highlighting the superior performance of ensemble learning in deciphering complex Android app behaviors. Beyond numerical advancements, our work holds practical significance in bolstering user security and privacy. This research marks a paradigm shift, demonstrating that machine learning techniques like XGBoost and Gradient Boosting Classifier can outperform traditional deep learning methods, setting a new benchmark for accuracy in Android malware prediction and paving the way for future advancements in mobile security.