Description
The doctoral candidate will design, implement, and evaluate a federated graph neural network-based intrusion detection framework for industrial IoT networks. The work combines federated learning, graph neural networks, centrality measures, and complex network properties to detect cyberattacks and anomalies while preserving data privacy across heterogeneous industrial sites. The candidate will be hosted at CESI LINEACT in Strasbourg and INRS Canada, with access to industrial demonstrators and IoT platforms, and will use Python, PyTorch, TensorFlow, and PyG for implementation and experimental evaluation.
