Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/84633
標題: Application of Back-propagation Neural Network to Formulate Exercise Prescription for Taiwanese College Students
作者: Ching-Hua Chiu
Meng-Hsiun Tsai
Yung-Kuan Chan
Shih-Pei Chang
Yi-Wen Hung
Tzu-Lin Wong
關鍵字: Physical fitness;Hidden layer
出版社: Taichung, Taiwan :Graduate Institute of Sports & Health Management, National Chung Hsing University
Project: International Journal of Sport and Exercise Science, Volume 3, Issue 2, Page(s) 37-42.
摘要: 
In this study, we attempted to apply back-propagation neural network (BPNN) to formulate exercise prescriptions
for Taiwanese college students. The purpose was to realize a rapid and accurate estimation of the exercise prescription
for students. Three thousand college students of both sexes aged 19–24 participated in this study. Data on five physical
fitness test parameters were collected, including subjects’ age, body mass index (BMI), and performance in three
exercises: sit and reach, 1-minute bent-leg curl-ups, and running. The data were then randomly divided into two groups:
training samples (n = 1800) and testing samples (n = 1200). Next, BPNN was utilized to estimate the exercise
prescription level of the samples. The sample data was divided to examine the learning ability of BPNN. The BPNN
network structure for this study encompasses an input layer (5 units), a hidden layer (5 units), and an output layer (4
units). The learning rate of the BPNN was assumed to be 0.5, and its learning cycle consisted of 850 rounds. The results
indicated that the mean accuracy rate for estimating the prescription level was 93.22% for training samples and 92.38%
for testing samples. In other words, the mean relative error was 6.78% for training samples and 7.62% for testing
samples, both of which were within the acceptable range. These results indicate that applying BPNN to formulate an
exercise prescription is feasible. Furthermore, because it is rapid and accurate, BPNN could prove to be a better option
than manual assessment. Further, computer applications based on the BPNN technology can be developed to assist
teachers and coaches in formulating student exercise prescriptions, thus conserving the cost and labor that would
otherwise be required in the case of manual assessment.
URI: http://hdl.handle.net/11455/84633
Appears in Collections:第03卷 第02期

Files in This Item:
File SizeFormat
84608-2.pdf759.93 kBAdobe PDFView/Open
Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.