Deep learning in physical education teaching mode based on the concept of flipped classroom

This study aims to explore the application of deep learning in a sports teaching model based on the flipped classroom concept to enhance student engagement and teaching effectiveness. It constructs a personalized recommendation system based on the self-attention mechanism, integrating deep learning algorithms to provide customized learning paths and resources for sports teaching. Implemented in a flipped classroom, the system collects and analyzes student learning data to optimize the personalized learning model. The study is conducted in a university sports course, dividing students into an experimental group and a control group. The experimental group uses the flipped classroom model supported by the personalized recommendation system, while the control group follows the traditional teaching model. The results show that the prediction accuracy of the personalized recommendation system reaches 95.15%, significantly higher than other models (such as Convolutional Neural Network (CNN)). The surveys indicate that the experimental group outperforms the control group in classroom participation, learning satisfaction, and problem-solving abilities, with statistically significant differences (P < 0.05). This demonstrates that the system significantly improves learning outcomes. The study concludes that the personalized recommendation system based on the self-attention mechanism can effectively overcome the limitations of traditional teaching models. This study offers valuable practical solutions for innovation in sports teaching and promotes the deep integration of educational technology and sports instruction.