An Application of an Adaptive Educational Management Approach Based on Reinforcement Learning and Transformers in Knowledge Tracing

Abstract

In the context of rapid advancements in technology, personalized and intelligent education has become a critical focus in modern education systems. This paper proposes an innovative IPSO-Reinforcement Learning-ViLBERT framework for adaptive education management, aiming to improve knowledge state tracking and learning path recommendation in personalized learning systems. By integrating Reinforcement Learning with ViLBERT, which efficiently processes multimodal data, and optimizing parameters using Improved Particle Swarm Optimization (IPSO), the proposed framework enhances both the accuracy and efficiency of learning recommendations. This research is aligned with the Sustainable Development Goals (SDGs), particularly Goal 4: Quality Education, by providing a system that tailors educational resources to meet individual learning needs, ensuring inclusive and equitable education for all. Experimental results show that the IPSO-optimized model outperforms traditional optimization methods, demonstrating its potential in improving educational outcomes and advancing personalized education technologies.