類神經網路在運動科學界的相關探討

The Analysis of the Artifical Neural Network on the Sports Science

古國宏(Guo-hong Gu)

《休閒運動保健學報》 7期 (2014/07) Pp. 106-115

https://doi.org/10.6204/JRSHP.2014.07.10

摘要

類神經網路由於其有別於傳統迴歸分析之具備非線性之預測、分類與診斷能力，特別是倒傳遞類神經網路，故近年以廣泛的被應用於學術界。近幾年來，由相關運動科學文獻資料可發現，國內外運動科學界，已有應用類神經網路於標槍的飛行距離模擬預測、高爾夫最佳之揮桿學習策略、學生體適能預測、推估垂直跳關節力矩、推估拔河不同身體重心長度變化量及不同身體重心角度之蹬力、大專運動會聖火傳遞路徑的分析等等，故可知類神經網路可運用的範圍及方向很廣。但在分析的過程中需注意不同的模式架構、訓練、測試方式都會影響到結果，這也增加了分析時的不確定性，故需要透過豐富的經驗及大量試誤學習來減少誤差。

關鍵字

類神經網路 ； 倒傳遞類神經網路 ； 運動科學

並列摘要

Artifical neural network due to its different from the traditional non-linear regression analysis , and it's with the greater prediction, classification and diagnostic capabilities, especially in back-propagation neural network. It has been widely used in recent years in academia. In recent years, the relevant literature can be found in sports science. According to the relative domestic and foreign articles, the neural network has been applied in the javelin flight distance simulation predictions, the best golf swing learning strategies, student physical fitness forecast, joint torque estimated by vertical jump, position of center of gravity and ground force estimated in tug of war by neural network, analyzing college games torch relay route, etc. These results showed that the scope and direction of the neural network can be applied very widely. But during the analysis, researcher should pay more attention to the different model, training, testing methods will affect the results, and which may also increases the uncertainty in analysis, and therefore it would require a lot relative experience and trial and error learning to reduce errors.

並列關鍵字

artifical neural network ； back-propagation neural network ； sports science