An investigation of the suitability of Artificial Neural Networks for the prediction of core and local skin temperatures when trained with a large and gender-balanced database

详细信息    查看全文

• 作者：K. Michael^a ; M.D.P. Garcia-Souto^b ; ^{p.garciasouto@ucl.ac.uk} ; P. Dabnichki^c ; ^{peter.dabnichki@rmit.edu.au}
• 关键词：ADALINE ; adaptive linear element ; AI ; artificial intelligence ; ANN/NN ; Artificial neural network(s) ; BMI ; body mass index (body mass of the subject divided by the square of the height in m ; i.e. kg/m2) ; Epoch ; a full data cycle ; i.e. each time the network is presented with a new training pattern ; FNN ; fuzzy neural network(s) ; Logsig ; the logistic sigmoid function ; MSE ; mean squared error ; Nodes ; the interconnections between hidden layers in the neural network (equivalent to neuron connections)
• 刊名：Applied Soft Computing
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：50
• 期：Complete
• 页码：327-343
• 全文大小：4462 K
• 卷排序：50

文摘

Neural Networks were created to predict core and local skin temperatures using a large gender-balanced experimental dataset. NNs significantly increase accuracy with respect to multi-linear regression models (e.g. R-value increase 81% for core). Core temperature is not practical neither required for the prediction of skin temperature using Neural Networks. The addition of Average clothing as an input is beneficial for the prediction of core, forehead and hands temperature. The best predictive models were found for skin temperature at hands and knees.