BruFence: Scalable machine learning for automating defense system
You are here: Home \ Projects \ BruFence

The aim of the BruFence project is to design systems based on machine learning and big data mining techniques that allow sensible and secure systems to automatically detect attacks and fraudulent behaviors.


From Wikimedia (Lotus Head, Johannesburg)

Advanced persistent threat detection market as well as scalable fraud detection is expected to have a high progression rate in the next years, therefore the project is twofold and addresses researches on:

  • Automatic detection of threats and attacks against communication systems, managed file transfer and collaboration platforms;
  • Automatic detections of fraud in large amount of transactions.

During the compilation of a risks and threats analysis, it is not trivial for defenders to foresee all the potential risks of a communication system. Recent attacks highlight the limits of current threats models and risks analysis methodologies. The aim of our research is to enhance the security of communication systems by enabling automatic learning from past and current attack attempts. The system will also be designed to report and assist the managers of the communication platform.

In the framework of fraud detection, automatic systems are essential since it is not possible or easy for a human analyst to detect fraudulent patterns in transaction datasets, often characterized by a large number of samples, many dimensions and online update. The research undertaken in the framework of the BruFence project will propose new efficient techniques for automatic detections of fraud in large amount of transactions.

Both application domains will take advantage of the research activities of BruFence aiming at developing a real-time framework that is able to compare in parallel a large number of alternative models in terms of nature, features, supervised or unsupervised data, predictive methods, scalability and quality criteria.

We will also investigate the exploitation of network data (social networks, communication networks, etc.) for fraud, privacy and security purposes. The rationale is that network connectivity provides information that can improve the accuracy of the prediction model.


From Wikimedia (John M. Kennedy T.)

The BruFence research will bring societal and economical value by addressing the urgent needs of the security community that finds itself facing the current evolutions of attacker techniques and the ineffectiveness of security based on static rule engines (antivirus, firewalls, IDS, application proxies…).

Projects Partners

Projects Sponsors

Logo Sopra Steria .jpg