样本

Using the genetic programming approachs as the modelling tool, all 17 samples were divided into a training set ( 15 samples) and a testing set ( 2 samples).

运用遗传程序设计方法作为建模工具，将全部17个样本（17年的调查数据）分为训练集（15个样本）、预测集（2个样本）。

In the model experiment, intact members are replaced by the prefabricated damage members to form damage structure. For test own limitations, some damage samples can be obtained by the model experiment and these damage samples can be considered as the testing samples of the neural networks.

在模型试验中，通过预制好的损伤构件来替换完好的构件形成损伤结构，由于试验本身的限制，部分损伤样本通过模型试验获得，所获得的损伤样本作为网络的测试样本。

The BP neural network is applied in face recognition. A face recognition model is established, and three training strategies i.e. sampling training, batch training and integrity training are studied.

将BP神经网络用于人脸识别，建立了人脸识别模型，研究了样本采样训练、样本批量训练和样本完整训练三种训练策略对识别率的影响。