(2025年11月4日) SI センター協力セミナー: Pierre-François Gimenez 氏，Franco Terranova 氏ご講演のお知らせ
2025.10.06
SI センター協力のもと、Pierre-François Gimenez 氏，Franco Terranova 氏をお招きしてセミナーを開催します。
開催日は2025年11月4日、時間は15時からの約1時間を予定しています。会場は本郷キャンパス工学部2号館242講義室です。
—
Title: Towards more realistic honeypots with synthetic network traffic injection
Abstract: Honeypots and honeynets need to be realistic to attract and
convince attackers to reveal their techniques. Several work on
realistic file systems, but realistic local network communication is
still an open question. In this work, we propose to generate synthetic
network traffic using generative machine learning techniques and
inject it into the network. This presentation entails some recent,
ongoing work on this subject.
Bio : Pierre-François Gimenez is a research scientist at Inria. He is
a member of the IRISA laboratory in the Protection of Information and
Resistance to ATtacks (PIRAT) team, where he works on network
intrusion detection and synthetic network data generation.
Applications include monitoring embedded systems, radio communication,
IT network communications, and computer systems. He holds a Ph.D.
Degree in artificial intelligence from IRIT, France. He is leading the
associate team SecGen between Inria and CISPA focused on generating
synthetic security data for intrusion detection system assessment, and
he is a member of French national projects on supervision (Superviz),
malware analysis (DefMal), and vulnerability search (REV).
Website: https://pfgimenez.fr/
—
—
Title: Modeling Attacker Behavior with RL Agents to Anticipate
Critical Vulnerability Paths
Description: In this talk, I will present the key contributions of my
PhD research on a proactive approach to cyber-defense: simulating
attacker behavior with reinforcement learning (RL) agents to
anticipate and identify the most critical vulnerability paths in
networks guided by a desired threat model. These agents navigate the
network by selecting vulnerabilities, aiming to uncover the most
critical attack paths.
The talk focuses, in particular, on two main contributions: (1)
Continuous CyberBattleSim—our extension of the Microsoft’s
CyberBattleSim simulator—supporting more realistic scenarios of
vulnerability allocations and cyber-terrain, with automated scenario
transitions, enabling more effective training and evaluation of RL
agents; (2) Strategies for improving the generalization and
scalability of RL agents by reformulating their observation and action
spaces. This includes leveraging embedding spaces with language models
for vulnerability representations and graph neural networks for
network topology representations.
Through these contributions, we move toward more scalable and
generalizable AI-driven defense solutions that anticipate attacker
behavior and strengthen the resilience of dynamic network
environments—for example, by using predicted attack paths to
prioritize vulnerability patching.
Bio: Franco Terranova is a 3rd-year Ph.D. student in Computer Science
at Université de Lorraine and INRIA (Nancy, France), where his
research focuses on deep reinforcement learning (RL) for cyber-attack
path prediction. During a visiting research period at the University
of Waterloo (Waterloo, Canada), he also worked on multi-agent RL for
secure and efficient virtual machine placement in an attacker-defender
turn-based framework.
He earned his Master’s degree in Artificial Intelligence and Data
Engineering from the University of Pisa (Italy) in 2023. His research
career has started exploring deep learning and RL applications during
research periods at the European Space Agency (Cologne, Germany) and
the Fermi National Accelerator Laboratory (Chicago, USA), including
work on RL for a self-driving telescope.
Website: https://terranovafr.github.io/