A Deep Learning Model to Predict Information Technology Students Learning Style in Adaptive Learning Systems

A Deep Learning Model to Predict Information Technology Students Learning Style in Adaptive Learning Systems

연구 분야: Artificial Intelligence

학회: 2025 4th International Conference on Sentiment Analysis and Deep Learning (ICSADL)

The personalization of learning and the diversity of learning styles among students have required pedagogical innovations in educational institutions. The emergence of disruptive technologies in education has accelerated the potential for course delivery dynamics in both online learning and traditional classrooms. Students in the 21st century possess distinct strengths and limitations and favor instructional methods that will effectively improve their academic performance. In the previous study where supervised learning algorithms were implemented, we achieved a relatively low accuracy. In this study, we proposed a deep learning Multilayer Perceptron (MLP) algorithm to predict the learning style of students using thirty-seven training set features. We employed the Synthetic Minority Over-sampling Technique (SMOTE) to augment the synthetic data for the minority class and achieve dataset balance when constructing the classifier. The hyperparameter of the MLP algorithm was optimized by multiple tests with varying training and testing split ratios to enhance the algorithm's predictive accuracy. The accuracy results show an improvement of the MLP deep learning classifier over the ensemble AdaBoost (RF) algorithm by 36.2%. The other performance indicators, such as the F-measure and Receiver Operating Characteristic (ROC) for the MLP, outperform the ensemble classifier and supervised learning techniques.