Research on Artificial Intelligence-Assisted Personalized Learning Path Generation Algorithm Based on Deep Learning

This study presents deep learning models' design, implementation, and evaluation to generate personalized learning paths in educational environments. Data was collected from diverse sources, including educational databases, learning management systems, user feedback, and surveys, to comprehensively understand student behaviours and preferences. The data was preprocessed through data cleaning, normalization, feature engineering, and augmentation to ensure quality and consistency. Extended Short-Term Memory networks and Transformer models were chosen for their abilities to handle temporal sequences and capture long-term dependencies in student learning behaviours. The models were trained systematically with Nvidia Tesla Volta V100 Graphics Processing Units. The generated learning paths were highly relevant and personalized to each student's unique context. The dynamic adaptability of LSTM and Transformer architectures allowed the models to provide personalized, data-driven recommendations supporting immediate learning needs and long-term growth. Future research will focus on enhancing real-time personalization through feedback mechanisms and expanding the scope to include multimodal data, ultimately contributing to more effective, scalable, and personalized learning systems.