REFORMING PHYSICAL EDUCATION TEACHING MODELS USING NEURAL NETWORK ALGORITHMS DRIVEN BY BIG DATA ANALYTICS

Abstract

BD (big data)-driven sports precision teaching can turn the exploration of causality to the discovery of correlation, pay attention to finding correlation in different data, and pay attention to exploring the law of correlation in the process. As a practical course, most of physical activities are the main courses, and different events have different technical action essentials. Therefore, this paper studies the reform of PET (physical education teaching) mode based on BPNN (Neural network) algorithm driven by BD. The theory of BPNN is applied to the TQE (Teaching quality evaluation) system of college PET, and the changing adaptive learning rate is adopted, which makes the network training automatically set different learning rates at different stages. In order to ensure the reliability of the application of neural network in college physical education TQE, GA (genetic algorithm) is introduced into the neural network to improve and optimize the network weight. It is found that the error values obtained by BPNN improved by GA are the best. Compared with other algorithms, the average error is reduced by 1.687%. The experiment proves that the application of GA-improved BPNN model in teaching evaluation is scientific, objective and reasonable.