Malware Detection Using RNA Encoding and Convolutional Neural Networks on the Malicious Network Dataset.

Area of Science:

• Cybersecurity
• Machine Learning
• Network Security

Background:

• Malware detection in network traffic is a critical cybersecurity challenge.
• Signature-based methods struggle with novel and polymorphic threats.
• Anomaly-based methods can detect new incursions but often yield high false positives.

Purpose of the Study:

• To propose a combined malware-detection framework utilizing RNA encoding and Convolutional Neural Networks (CNNs).
• To develop distinct and integrated detection modules for comprehensive threat identification.

Main Methods:

• A framework integrating RNA encoding of network-flow attributes with CNN classifiers.
• Three functionalities: Signature-CNN for known threats, Anomaly-CNN for unknown threats, and Hybrid-CNN combining both.
• Supervised learning on 10,000 samples with a 70/30 train-test split, using Python and deep learning libraries.

Main Results:

• Signature-CNN achieved 91% accuracy for known threats.
• Anomaly-CNN demonstrated a 93% detection rate for unknown malware.
• Hybrid-CNN delivered the best performance with a 95% detection rate and 94.5% F1 score.

Conclusions:

• RNA encoding combined with CNN classifiers provides a robust and scalable solution for network malware detection.
• The hybrid approach effectively addresses limitations of traditional signature-based and anomaly-based methods.