Disciplines
Other Social Sciences (40%); Electrical Engineering, Electronics, Information Engineering (20%); Computer Sciences (40%)
Keywords
-
Network Security,
Artificial Intelligence,
Graph Neural Networks,
Network Measurements
In today`s digital age, the rise of cybercrime poses a significant threat to our online security. The GRAPHS4SEC research project aims to revolutionize the way we protect our networks using cutting-edge technology known as Graph Neural Networks (GNNs). Traditional methods of using Artificial Intelligence (AI) and Machine Learning (ML) for network security have fallen short. They struggle to adapt, perform poorly in real-world situations, and are susceptible to cyber-attacks. The primary reason for these limitations is the lack of specialized AI/ML technology designed specifically for network security challenges. At GRAPHS4SEC, the spotlight is on harnessing the potential of GNNs to enhance cybersecurity. These are advanced systems that excel at understanding and learning from interconnected information, perfect for the relational nature of network security data. Imagine them as digital detectives for our online safety. The project has three main goals: Smart AI-powered Cybersecurity Algorithms: we will explore new ways to model and learn from network security data using graph-based approaches. This means creating smart algorithms that understand the complex relationships within cybersecurity information. Understanding the Added Value: the team will compare how well GNN-based approaches perform against traditional AI/ML methods. This involves evaluating detection capabilities, adaptability, scalability, and resilience against cyber threats. It`s like putting the new technology to the test to see if it outperforms the old. Real-World Cybersecurity Applications: the ultimate aim is to apply GRAPHS4SEC technology in real-world scenarios. For that, will focus on four critical areas of cybersecurity, with a special emphasis on the detection and early mitigation of phishing and malicious websites, a pervasive and significant threat that poses widespread harm in today`s digital landscape. In essence, GRAPHS4SEC strives to create a new generation of powerful, resilient, and effective AI-driven cybersecurity tools. By harnessing the unique capabilities of GNNs, the project aims to enhance our ability to safeguard against cyber threats and make the online world a safer place for everyone.
- Stefano Secci, Conservatoire National des Arts et Metiers (CNAM) - France
- Andrzej Duda, Université Grenoble Alpes - France
- Maciej Korczynski, Université Grenoble Alpes - France
- Marco Mellia, Politecnico di Torino - Italy
- Franco Scarselli, Università di Siena - Italy
- Pere Barlet-Ros, Universitat Polytecnica de Catalunya - Spain
- Kimberly Claffy, University of California San Diego - USA
- Hamed Haddadi, Imperial College of London
Research Output
- 3 Citations
- 8 Publications
-
2024
Title Timeless Foundations: Exploring DC-VAEs as Foundation Models for Time Series Analysis DOI 10.23919/tma62044.2024.10559129 Type Conference Proceeding Abstract Author González G Pages 1-4 -
2024
Title Certainly Uncertain: Demystifying ML Uncertainty for Active Learning in Network Monitoring Tasks DOI 10.23919/cnsm62983.2024.10814433 Type Conference Proceeding Abstract Author Dietz K Pages 1-7 -
2025
Title Do We Really Need Reference-Based Phishing Detection? Unleashing the Power of GNN DOI 10.23919/tma66427.2025.11096997 Type Conference Proceeding Abstract Author Song T Pages 1-4 -
2025
Title FREKit: A Flexible Simulator for Probabilistic Congestion and Rerouting in Multi-Protocol Networks DOI 10.23919/tma66427.2025.11096995 Type Conference Proceeding Abstract Author Vanerio J Pages 1-4 -
2025
Title Towards Intelligent Resource Allocation in Highly-Distributed Content Delivery Networks Using Graph Neural Networks DOI 10.23919/tma66427.2025.11097019 Type Conference Proceeding Abstract Author Vanerio J Pages 1-4 -
2025
Title TSGFM - Towards a Graph Foundation Model for Time Series Analysis in Network Monitoring DOI 10.23919/tma66427.2025.11096996 Type Conference Proceeding Abstract Author Latif-MartÃnez H Pages 1-4 -
2025
Title TSGFM - Graph Neural Networks for Zero-Shot Time Series Forecasting in Network Monitoring DOI 10.23919/cnsm67658.2025.11297447 Type Conference Proceeding Abstract Author Latif-MartÃnez H Pages 1-9 -
2025
Title Malicious Domain Names Detection with DeepDGA, a Hybrid Character and Word Embeddings Deep Learning Architecture DOI 10.23919/cnsm67658.2025.11297537 Type Conference Proceeding Abstract Author Aravena L Pages 1-6