Adaptive learning systems facilitate the personalization of education using Artificial Intelligence (AI) that analyzes student data so as to align content and support. By using Deep Learning (DL) models and reinforcement learning, the system keeps changing the learning paths. In such a way, the system ensures the scalability, responsiveness, and effectiveness of education, which varies according to various profiles of learners. The proposed system, in this study, uses the Open University Learning Analytics Dataset (OULAD) and incorporates the reinforcement learning (RL) component of personalization in learning. The system in question starts with preprocessing and feature extraction, in which the key behavioral, demographic, and performance indicators are isolated. K-means clustering is used to divide students into intervention-oriented profile. DQN RL model suggests interactive learning that are most appropriate to the current academic conditions and engagement of the evolving learner, at this time or that. The predictive ability of the system is rather high as the evaluation shows the test accuracy of 0.9991, the precision of 0.9992, the recall of 0.9984, and the F1-score of 0.9988. These facts therefore verify the validity of the model and how it can be generalized to other kinds of learners. Thus, it can design an inclusive, scalable, and effective learning environment by continuously changing with every behavioral change and academic status of a student.
Liang et al. (Sat,) studied this question.