What is it about?
One increasingly popular approach of AI-based network security models is using Graph Neural Networks. However, most academical proposals miss some of the practical difficulties involving practical deployment of such models. We put forth a pipeline that is both highly performant while guaranteeing practical implementability.
Featured Image
Photo by GuerrillaBuzz on Unsplash
Why is it important?
Our work provides a necessary contrast to the quickly growing number of papers proposing GNN-based Network Intrusion Detection system, by rebasing our efforts to actual, practical problem solving instead of theoretical evaluations.
Perspectives
Writing this paper has been exciting and challenging, with it being an extension of my master's thesis and thereby my first publication. Though still quite removed from an actual practical graph-based, I hope researchers get inspired by our emphasis on practicality and are convinced by the our experiments showing the importance of large-scale pretraining.
Louis Van Langendonck
Universitat Politecnica de Catalunya
Read the Original
This page is a summary of: PPT-GNN: A Practical Pretrained Spatio-Temporal Graph Neural Network for Network Security, June 2025, Institute of Electrical & Electronics Engineers (IEEE),
DOI: 10.1109/eurospw67616.2025.00026.
You can read the full text:
Contributors
The following have contributed to this page
