【摘要】 本文将神经网络模式识别用于金属间化合物3元填隙d88结构形成条件的判别;神经网络由3层组成,训练用后向传播算法。为了评价所得模型的行为,使用了交叉验证法。计算结果表明,当适当的键参数作为输入时,71种化合物能被正确地分类,且隐层节点数经优选后可以提高分类的准确率,减少计算时间。更多还原

【Abstract】 In this paper, neural networks pattern recognition was applied to discri(?)inate the formation conditions of ternary chinse d88 structure of internal metal compound(?). The neural network was constituted of three layers and the back propagation algorithm was used. To evaluate the performance of the obtained model the cross-validation strategy was employed. The results showed that 71 compounds could be classified satisfactorily by neu(?)d networks when the bond parameters were used as the input, and there was an optimal number of hidden layer nodes where the percent of correct classification is higher while the calculi tion time used is fewer.更多还原