<![CDATA[This study proposes an optimized Artificial Neural Network (ANN) framework for the automatic classification of student learning styles based on the Visual, Auditory, and Kinesthetic (VAK) model to support educational personalization. The system utilizes data from a 36-item questionnaire, which are preprocessed using z-score standardization and encoded into numerical features. Several ANN configurations were evaluated to examine the influence of network depth, activation functions, regularization strength (λ ranging from 1 × 10⁻⁴ to 1 × 10⁻²), and the number of training iterations between 500 and 2000. Model training employed the Adam optimizer combined with early stopping to ensure stable and efficient convergence. The results demonstrate that proper data standardization and suitable regularization significantly enhance model generalization and training stability. The optimal model consists of two hidden layers with 64 and 32 neurons using ReLU activation and λ = 1 × 10⁻², achieving an accuracy of 91.9% and a macro-F1 score of 0.908. Overall, systematic hyperparameter optimization improves the robustness and reliability of ANN-based learning style classification for adaptive learning systems.]]>
Copyrights © 2026
