【摘要】 多层神经元网络中隐含层的层数及其每一层的神经元数的多寡 ,至今仍无法用一个解析式精确求得 .提出了一种基于黄金分割法的算法 ,用于求解三层前馈神经网络中的隐层节点数 ,所得结果有助于提高三层前馈神经网络整体品质 .仿真结果表明该算法能较快地找到较优的隐层节点数更多还原

【Abstract】 So far there is still no analytic formula that can precisely solve the quantity of the neural elements and the numbers of hidden layers in the multi-layer neural networks. The author in this essay put forward an algorithm based on the golden section to solve the numbers of hidden nodes in three-layer BP network. The outcome is conductive to improve the whole quality of three-layer BP network.The result of simulation indicates that people can find the suitable quantity of hidden layers nodes quickly using the algorithm.更多还原