The situation and transformation of sports governance in national colleges and universities under the background of double reduction

Abstract

This paper aims to improve the quality of sports development by using particle swarm optimization (PSO) based long and short-Term memory network (LSTM) and support vector machine (SVM) based sports performance prediction model (SPPM) to realize the shift to information technology (IT)-based sports governance in the context of the "Double Reduction". The effects of LSTM and PSO-LSTM are evaluated by testing individual exercise physiological data like heart rate and blood oxygen values. Test the effectiveness of particle swarm optimization neural network's sports performance prediction model with sample data and conduct comparative experiments with multiple linear regression, genetic algorithm optimization BP neural network, and firefly optimization BP neural network, respectively. Analyzing the characteristics of the extracted samples and performing a paired-sample t-Test on the sample size revealed the gap between service perceptions and expectations. The analysis shows that the overall expectation mean of parents is 4.482, and the actual perceived mean is 3.966. All the indexes fail to reach the ideal level of parents, which indicates that the delayed sports service proposed by the institutional sports needs to be further adjusted.