Learning Mechanisms for Detecting Abnormal Behaviors in Embedded Systems

This doctoral position at the Laboratoire des Systèmes Embarqués Sécurisés (LETI) focuses on addressing security vulnerabilities in critical infrastructures where embedded systems are widely deployed. The research centers on developing dynamic intrusion detection systems (IDS) using machine learning anomaly detection that can adapt over time without catastrophic forgetting. The successful candidate will explore re-learning strategies, reinforcement learning, and federated learning paradigms to continuously update models on evolving system behaviors. The project is hosted in France and offers an immersive research environment within a prominent national research organization.

• Design and implement machine learning models for anomaly detection in embedded systems
• Investigate re-learning mechanisms to update model knowledge without losing prior historical data
• Explore alternative learning paradigms such as reinforcement learning and federated learning
• Evaluate system performance and robustness against malicious attacks in critical infrastructures
• Publish research findings in peer-reviewed conferences and scientific journals