A Reference Architecture for Personalized Cognitive Education via Hierarchical Federated Learning

In this paper, we propose a conceptual reference architecture specialized for multi-level personalized learning in digital education systems that utilizes hierarchical federated learning. The usage of different machine learning based models by educational institutions faces challenges due to the lack of sufficient training data, heterogeneous learner profiles, and strict privacy regulations. The proposed architecture addresses these by organizing the institutions into multiple hierarchical levels that leverage federated learning. The architecture also provides high availability and scalability, while still complying with stringent data-privacy requirements.