The increasing digitalization of education introduces both opportunities and significant challenges. One of the key challenges is the need to personalize learning experiences while safeguarding student data privacy. The proposed system leverages Federated Learning to ensure that sensitive information remains on local devices instead of being centralized, thus adhering to modern data protection standards.
Using a Deep Neural Network and validated against the ASSISTments dataset, the study identifies FedProx as a superior aggregation strategy over standard methods. This federated approach not only achieves a competitive F1-Score of 76.28%, which is 82.85% of a centralized model’s performance, but it also highlights the potential for scalable and secure personalization strategies in educational technology. The findings underscore the importance of developing methodologies that respect privacy without compromising the quality of educational experiences.
👉 Pročitaj original: arXiv AI Papers
