Performance Paradox: Evaluating Machine Learning and Deep Learning Strategies for Enhanced Mobile Security
Kausal SD, Saranya G · 2025
With the rapid expansion of mobile device usage and the increasingly sophisticated nature of cyber threats, mobile security has become critically important. Mobile malware poses significant risks by targeting sensitive data and exploiting system vulnerabilities. This study presents a comparative analysis of ten machine learning and deep learning models aimed at improving mobile malware detection and overall mobile security. The models were tested on a dataset of 15,036 mobile applications, comprising 5,560 malware samples and 9,476 benign applications, sourced from real-life data on Datahub. Among the evaluated models, the Deep Neural Network (DNN) achieved the best performance, with an accuracy of 99.40 percentage. These findings underscore the potential of DNN for real-time mobile security applications, enabling effective detection and prevention of malicious activities on mobile platforms.