Identifying Mobile Malware Attacks on Android Devices Using Machine Learning Algorithms Identificación de ataques de Malware móvil en dispositivos Android mediante algoritmos de aprendizaje automático

Abstract

Objective: to identify mobile malware attacks on Android devices using machine learning algorithms. Methodology: based on the KDD (Knowledge Discovery in Databases) process, a structured approach that organizes data mining in clearly defined stages. This model ensures precise control at each stage, allowing for efficient information extraction, transformation and analysis. Results: The LightGBM algorithm demonstrates, through the confusion matrix, its ability to efficiently process large volumes of data and multiple features, contributing to more accurate classification. In addition, this model excels in evaluation metrics, achieving optimal performance compared to other machine learning approaches. Discussion: Machine learning is key in cybersecurity, improving the detection of threats such as malware and DDoS attacks. LightGBM stands out for its efficiency achieving the best accuracy (94.4%), followed by XGBoost with high performance, but longer computation time. Random Forest, although faster, presents lower accuracy. Conclusions: Machine learning has revolutionized cybersecurity, strengthening the detection of threats such as malware and intrusions. In identifying mobile malware on Android, LightGBM stood out for its accuracy and efficiency in handling unbalanced data, outperforming other models in key metrics. However, the challenge remains balancing accuracy and resource consumption, especially on mobile devices with limited hardware.