Application of Neural Network in Prediction of Brassiere-Wearing Effect

Brassiere as a foundation garment functions not only to protect woman\´s breast, but also to provide formative beauty. In order to meet the individualized need of customers and serve for apparel e-commerce, a prediction model of brassiere-wearing effect based on BP neural networks was established in this research. Firstly, the influencing factors of bust shape change were investigated, including the principal components of naked bust measurements and the structural parameters of brassiere production. Then by using the Matlab neural network toolbox, a back-propagation artificial neural network model was created. It consisted of 6 subnets which could predict the change ratio of a given effect parameter. The function "postreg" was applied to determine the final architecture of each subnet. Finally, it was validated that the predicted result of this model showed satisfying effects.