Deep Ensemble Neural Network Classifier for Android Malware Detection

Authors: P Sumalatha and G.S. Mahalakshmi, Anna University, India

Key Points:

  • Malware poses a significant threat to computer security.
  • AI techniques, particularly machine learning (ML) and deep learning, are gaining traction in malware detection.
  • Existing deep learning methods rely on a single learning model.
  • This paper proposes a Deep Ensemble Framework (DEF) for improved malware detection on Android devices.
  • DEF utilizes two analysis techniques:
    • Convolutional Neural Network (CNN) to analyze a grayscale image generated from the malware.
    • Long Short-Term Memory (LSTM) to analyze the opcode sequence of the malware.
  • The results are then combined using a stacking ensemble approach for better classification accuracy.
  • The framework is trained on malware samples collected from various sources.

Conclusion:

The proposed DEF framework demonstrates improved performance compared to existing methods in terms of speed and accuracy for Android malware detection.

Further Studies:

The authors suggest potential areas for further research:

  • Explore different ensemble techniques for potentially better performance.
  • Investigate the effectiveness of DEF against emerging malware variants.
  • Analyze the impact of including additional features for analysis.

This summary captures the essence of the research paper, highlighting the problem, proposed solution, key findings, and opportunities for future exploration.

Published
Categorized as Blog