Improved Understanding of Malware Detection Using LEN

Inspiration

In today's ever evolving Cybersecurity landscape, Malware Detection is a constant challenge, due to the sheer volume of new threats. Traditional signature-based approaches struggle to keep pace, while Machine-Learning approaches face issues of generalization to unseen samples, and a lack of explanation for the instances identified as malware, which is crucial for trust and legal compliance.

Our Solution

Our solution for the problem in question consists of using Logic Explained Networks (LENs), a recently proposed class of interpretable neural networks that generate explanations in the form of First-Order Logic (FOL) rules. Our project demonstrates that LENs not only outperform the traditional models but also rival the performance of black-box models, all while providing human-understandable explanations.

Conclusion

By leveraging the RaDaR dataset, we can show that LENs can enhance malware detection effectiveness and transparency, making them a viable and promising solution for building people's trust in AI-driven Cybersecurity systems.

Built With

• graphdb
• java
• owl
• protege
• rdf
• sparql