Imagine a group of detectives who want to share clues without actually showing each other their secret files to protect privacy. This system lets them train a smart AI together that spots patterns of bad behavior across the web and police records. It's like a collective brain that finds the bad guys without ever risking the leak of sensitive personal data.
Law enforcement agencies struggle to analyze massive amounts of data from the web and internal databases without violating privacy laws. This leads to inefficient resource allocation and missed opportunities to stop cybercrime and terrorism.
A decision-support system integrating Federated Learning and User and Entity Behavior Analytics (UEBA) to monitor social networks and police databases.
If you are a threat intelligence provider dealing with fragmented data across different jurisdictions — this project developed a federated learning system that identifies criminal communities while keeping data private. This allows for better detection of cybercrime patterns without violating strict privacy laws.
If you are a police department dealing with limited manpower and complex cyber-crime — this project developed a decision-support system that analyzes social networks and the deep web. It helps you allocate your resources to the most critical areas by identifying high-risk users.
If you are an AML software firm dealing with the need to analyze behavior across different banks without moving sensitive data — this project developed User and Entity Behavior Analytics (UEBA) using federated learning. This ensures ethical data processing while spotting suspicious activity patterns.
Based on available project data, the EU contribution is EUR 5,338,197, but the commercial price for the final software is not listed.
The project uses Big Data and AI techniques designed to monitor social networks and the deep web, suggesting a capacity for large-scale data processing.
Based on available project data, specific licensing terms are not provided; however, it is a Horizon-IA project involving 14 partners.
The system is built to be privacy-preserving and ethical, specifically using Federated Learning to keep data distributed rather than centralized.
The project period runs from 2024-09-01 to 2027-08-31.
The consortium is well-balanced for commercialization, featuring 14 partners across 9 countries. With 6 industry partners (including 7 SMEs total) and an industry ratio of 43%, there is a strong focus on practical application rather than just academic research. The mix of 2 universities and 3 research centers ensures the AI and Federated Learning components are grounded in science while being driven by market-ready goals.
Contact Fundacion Centro Tecnoloxico de Telecomunicaciones de Galicia
Contact us to track the development of this privacy-preserving AI platform for security applications.
