【摘要】 讨论了人工神经网络中最常用的多层前馈网络 ( BP网络 )及误差反向传播算法应用于化学和环境科学时要考虑的几个问题 :网络的输入与数据的归一化 ;隐含层数、隐含层节数和学习速率 ;训练集与监控集 ;网络误差 ;初始权重更多还原

【Abstract】 The problems about the Back Propagation (BP) network applied to chemistry and environmental science are discussed in this paper Those are 1 input of network and the scaling of data, 2 the number of hidden layer, hidden nodes and the learning rate, 3 the train set and the test set, 4 network errors, 5 The weights更多还原