The company specializes in aggregating and analyzing cybersecurity event logs from diverse sources, including computer networks, devices, and intelligence providers. The current challenge is filtering through vast amounts of unstructured data to detect and predict cyberattacks effectively while minimizing false positives—a critical issue in the industry often referred to as "alert fatigue."

The proposed R and D project seeks to develop a machine learning framework capable of recognizing and clustering patterns of malicious activities within large volumes of low-fidelity detection signals. Inspired by methodologies used in other domains like healthcare and finance, the research will focus on the correlation of disparate events into coherent attack narratives.

The ideal consortium would include:

- Academic partners with expertise in AI/ML for developing innovative algorithms (e.g., deep learning, anomaly detection, graph neural networks).
-Cybersecurity specialists to ensure domain-relevant feature selection, synthetic data generation, and robust model validation.
Optionally, an industrial partner (e.g., managed security service providers or tech vendors) for strategic integration, data enrichment, and market research.
The project aligns with the Horizon Europe call DIGITAL-ECCC-2024-DEPLOY-CYBER-07, with an anticipated budget of €5 million. Expressions of Interest (EOIs) are open until May 31, 2025.
Advantages and innovation
The project offers a transformative approach to cybersecurity by:

- Applying cutting-edge AI techniques like deep learning and graph neural networks to a high-impact domain.
- Reducing the industry s reliance on noisy data by focusing on actionable insights and precise detections.
- Bridging gaps in current cybersecurity solutions through interdisciplinary collaboration.
- While the initial focus is on cybersecurity, the methodologies developed are expected to have cross-domain applicability, benefiting industries such as healthcare, finance, and social sciences.
Technical Specification or Expertise Sought
The consortium seeks:

- Academic institutions specializing in machine learning, AI, and cybersecurity research for algorithm development and proof-of-concept studies.
- Cybersecurity-focused entities to provide domain expertise and assist in feature validation, model testing, and real-world simulation.

Optionally, industrial partners (e.g., MSSPs or technology vendors) for integrating the developed solutions into existing cybersecurity frameworks.
Partners should be experienced in EU research projects and committed to collaborative, interdisciplinary development.

Framework program
Horizon Europe
Call title and identifier
DIGITAL-ECCC-2024-DEPLOY-CYBER-07 (deadline 2024/01/21) but open to alternatives
Submission and evaluation scheme
Anticipated project budget
EUR 5.000.000
Coordinator required
No
Deadline for EoI
31/05/2025
Deadline of the call
31/05/2025
Project duration in weeks
Web link to the call
Project title and acronym
Identification of Attack Flows from Low-Fidelity Detection Signals Using Machine Learning