Personalized Teaching Strategy Using Convolutional Neural Networks and Clustering Algorithms for Enhanced Performance and Efficiency

In a large-scale teaching environment, it is difficult for teachers to make effective personalized adjustments based on the different characteristics of each student. Taking basketball as an example, this study applies a personalized teaching strategy that integrates Convolutional Neural Network (CNN) and Ordering Points To Identify the Clustering Structure (OPTICS) clustering algorithm: students’ physical fitness, technical data and learning progress data are collected, and CNN is used to extract students’ shooting posture, step control and other sports characteristics. Then, Singular Value Decomposition (SVD) is used to decompose the data matrix into three low-dimensional matrices, which represent students’ sports characteristics, skill action weights and students’ preference information, thereby reducing redundant data and extracting highly recognizable features. The OPTICS algorithm is used to cluster the reduced-dimensional data, calculate the distance and local density between data points, and group them according to students’ technical level and ability. Finally, a personalized teaching strategy is formulated based on the clustering results. Real-time feedback on students’ progress is given during the teaching process, and the personalized teaching strategy is dynamically adjusted. The results show that the Normalized Discounted Cumulative Gain (NDCG) recommended by the model for technical learners is 0.80; the average field goal percentage is increased from 60% to 75%; the teaching efficiency ratio for beginners is 0.94. The method adopted can achieve high teaching accuracy and efficiency, and significantly improve the students’ basketball performance.