A Study on the Optimal Design of Reinforced Learning-Driven Personalized Physical Training Strategies in Physical Education Instruction DOI Open Access

Yinghui Jiang,

Duqian Ding

Applied Mathematics and Nonlinear Sciences, Journal Year: 2025, Volume and Issue: 10(1)

Published: Jan. 1, 2025

Abstract Reinforcement learning is applied to recommender systems balance the relationship between new and existing items improve accuracy of recommended items. In this paper, a personalized physical training strategy recommendation model combining LSTM reinforcement proposed perform optimize education process. A SOM neural network utilized segment fitness data different students. On basis, utilizes long short-term interest acquisition module obtain user’s real-time preference convert sequence processing problem into Markov decision Adding high low scoring decision-making actions pseudo-twin network, delayed rewards were obtained, noisy interaction records removed.The clustering method obtained characteristics each class students’ fitness, which paved way for subsequent individualized strategies. The in paper Ave_RMSE Ave_MAE values that outperformed other algorithms on two datasets. are 56.68% 68.70% higher than KNNWithMeans mL-1m dataset, respectively. There similarities dissimilarities strategies with recommendations experimental subjects. superiority based optimization has been demonstrated.

Language: Английский

A Study on the Optimal Design of Reinforced Learning-Driven Personalized Physical Training Strategies in Physical Education Instruction DOI Open Access

Yinghui Jiang,

Duqian Ding

Applied Mathematics and Nonlinear Sciences, Journal Year: 2025, Volume and Issue: 10(1)

Published: Jan. 1, 2025

Abstract Reinforcement learning is applied to recommender systems balance the relationship between new and existing items improve accuracy of recommended items. In this paper, a personalized physical training strategy recommendation model combining LSTM reinforcement proposed perform optimize education process. A SOM neural network utilized segment fitness data different students. On basis, utilizes long short-term interest acquisition module obtain user’s real-time preference convert sequence processing problem into Markov decision Adding high low scoring decision-making actions pseudo-twin network, delayed rewards were obtained, noisy interaction records removed.The clustering method obtained characteristics each class students’ fitness, which paved way for subsequent individualized strategies. The in paper Ave_RMSE Ave_MAE values that outperformed other algorithms on two datasets. are 56.68% 68.70% higher than KNNWithMeans mL-1m dataset, respectively. There similarities dissimilarities strategies with recommendations experimental subjects. superiority based optimization has been demonstrated.

Language: Английский

Citations

0