avatar

Hi, I’m Léo

I am a researcher at the SnT (Interdisciplinary Centre for Security, Reliability and Trust). I received my Ph.D. in 2024 from IMT Atlantique. Before that, I received an engineering degree in information security from ENSIBS and worked at Orange Cyberdefense as a part-time employee for three years.

Research

My research interests span different aspects of distributed systems, with a focus on network and system security, as well as decentralized and federated learning. During my Ph.D., I applied federated learning to the collaborative training of intrusion detection models, studying its limits in terms of data heterogeneity and resistance to Byzantine contributions. We proposed counters measures to ensure the quality of the global models, as well as a new way to generate datasets for distributed intrusion detection. I am now focusing on relationship extraction and inference under uncertainty in distributed systems, dealing with challenges of partial observability. I am also interested on the network optimization of decentralized learning in the context of constrained environments. Alongside my research, I taught topics related to network security to Master’s and Engineering students, worked on scientific popularization, and now contribute to the organization of national and international workshops and conferences.

Publications

International peer-reviewed journals

  1. Léo Lavaur, Yann Busnel et Fabien Autrel. “Investigating the Impact of Label-flipping Attacks against Federated Learning for Collaborative Intrusion Detection”. In: Computers & Security 156 (2025).
  2. Léo Lavaur, Marc-Oliver Pahl, Yann Busnel et Fabien Autrel. “The Evolution of Federated Learning-based Intrusion Detection and Mitigation: A Survey”. In: IEEE Transactions on Network and Service Management. Special Issue on Network Security Management (juin 2022).

International peer-reviewed conferences

  1. Léo Lavaur et Jérôme François. “WaterLLMarks: In-data User Tracing for Distributed LLMaaS Environments”. In: Proceedings of the IEEE 45th International Conference on Distributed Computing Systems Workshops (ICDCSW). Glasgow, UK: IEEE, juill. 2025.
  2. Léo Lavaur, Pierre-Marie Lechevalier, Yann Busnel, Romaric Ludinard, Géraldine Texier et Marc-Oliver Pahl. “RADAR: Model Quality Assessment for Reputation-aware Collaborative Federated Learning”. In: Proceedings of the 43rd International Symposium on Reliable Distributed Systems (SRDS). Charlotte, NC, USA, sept. 2024.
  3. Léo Lavaur, Yann Busnel et Fabien Autrel. “Systematic Analysis of Label-flipping Attacks against Federated Learning in Collaborative Intrusion Detection Systems”. In:Proceedings of the 19th International Conference on Availability, Reliability and Security (ARES). Workshop on Behavioral Authentication for System Security (BASS). Vienna, Austria, août 2024.
  4. Léo Lavaur, Yann Busnel et Fabien Autrel. “Demo: Highlighting the Limits of Federated Learning in Intrusion Detection”. In: Proceedings of the 44th International Conference on Distributed Computing Systems (ICDCS). Jersey City, NJ, USA, juill. 2024.

National peer-reviewed conferences

  1. Yann Busnel, Léo Lavaur, Pierre-Marie Lechevalier, Romaric Ludinard et Géraldine Texier. “Apprentissage fédéré cherche modèles de qualité, bonne réputation exigée”. In: ALGOTEL 2025: 27èmes Rencontres Francophones sur les Aspects. Saint Valéry-sur-Somme, France, juin 2025.
  2. Léo Lavaur, Pierre-Marie Lechevalier, Yann Busnel, Marc-Oliver Pahl et Fabien Autrel. “Metrics and Strategies for Adversarial Mitigation in Federated Learning-based Intrusion Detection”. In: Rendez-vous de la Recherche et de l’Enseignement de la Sécurité des Systèmes d’Information (RESSI). Neuvy-sur-Barangeon, France, mai 2023.
  3. Léo Lavaur, Benjamin Coste, Marc-Oliver Pahl, Yann Busnel et Fabien Autrel. “Federated Learning as Enabler for Collaborative Security between Not Fully-Trusting Distributed Parties”. In: Proceedings of the 29th Computer & Electronics Security Application Rendezvous (C&ESAR). Rennes, France, oct. 2022.

Tutorials

  1. Yann Busnel et Léo Lavaur. “Federated Learning and Network Security: Foundations, Potential, and Resilience”. In: Proceedings of the 45th International Conference on Distributed Computing Systems Workshops (ICDCSW). Glasgow, UK: IEEE, juill. 2025.
  2. Yann Busnel et Léo Lavaur. “Tutorial: Federated Learning × Security for Network Monitoring”. In: Proceedings of the 44th International Conference on Distributed Computing Systems (ICDCS). Jersey City, NJ, USA, juill. 2024.
  3. Yann Busnel et Léo Lavaur. “Federated Learning × Security for Network Management”. 15th International Conference on Network of the Future (NoF). Izmir, Turkey, sept. 2023.
  4. Yann Busnel et Léo Lavaur. “L’interêt de l’apprentissage fédéré dans le cadre de la détection d’incidents sur les réseaux et/ou systèmes à grande échelle”. Ecole de Printemps Recherche de l’EUR CyberSchool. Rennes, France, avr. 2023.

Non peer-reviewed conferences and posters

  1. Léo Lavaur, Marc-Oliver Pahl, Yann Busnel et Fabien Autrel. “Federated Security Approaches for IT and OT”. In: Journée thématique du GT sur la Sécurité des Systèmes, Logiciels et Réseaux (GT-SSLR). Mai 2021.
  2. Léo Lavaur. “Poisoning Attacks against Collaborative Federated Learning for Intrusion Detection”. Journées non-thématiques du GDR RSD. Jan. 2024.
  3. Léo Lavaur. “Trustable Federated Learning for Intrusion Detection in Heterogeneous Systems”. Colloque IMT “Sécurité et Résilience”. Prix du meilleur poster ex-æquo. Avr. 2023.
  4. Léo Lavaur. “Federated Approaches for Defending Cyber-Attacks”. Journées non-thématiques du GDR RSD. Jan. 2023.
  5. Léo Lavaur. “Federated Approaches for Defending Cyber-Attacks”. Colloque IMT “Gestion de crise et numérique: nouvelles menaces et nouvelles solutions”. Mars 2022.

Teaching

Practical sessions

MOOCs

Medias

Radio