Evaluating machine learning algorithms in detecting Android ransomware

  • Linet Momposhi

    Western Sydney University thesis: Master's thesis

    Abstract

    Ransomware attacks on Android devices have been increasing in recent years, posing a signif- icant threat to users’ data and privacy. This research presents a comprehensive evaluation of four popular machine learning algorithms – KNN, NN, Random Forest (RF), and SVM – in detecting Android ransomware. The study leverages a publicly available dataset from Kaggle, containing ten different types of ransomware attacks and benign instances of Android applica- tions, extracting relevant features for analysis. The performance of each classifier is assessed using various evaluation metrics, including accuracy, precision, recall, and F1-score. The results demonstrate that the RF classifier achieves the highest accuracy of 96.22%, followed by SVM with an accuracy of 83.51%, NN at 81.91%, and finally KNN at 70.49%. Furthermore, the research explores the strengths and limitations of each algorithm, providing insights into their suitability for real-world ransomware detection scenarios. The findings contribute to the devel- opment of robust and efficient security mechanisms for safeguarding Android devices against the evolving threat of ransomware.
    Date of Award2024
    Original languageEnglish
    Awarding Institution
    • Western Sydney University
    SupervisorAlana Maurushat (Supervisor) & Rodrigo Neves Calheiros (Supervisor)

    Cite this

    '