摘要
在应用多层前向人工神经网络解决实际问题时,经常存在的一个问题是,不能合理地选择神经网络中隐层神经元的数目.针对这一问题,给出一种在多层前向人工神经网络的学习过程中,通过合并、舍弃和转化的方式,优选神经网络中隐层神经元数目的方法,从而达到优化网络结构的目的.
During the application of multilayer feedforward artificial neural networks, there always exists a problem, that is the improper choice of the number of the hide nodes in the neural networks. This paper provides an optimum seeking method to choose the optimal number of hide nodes, by way of combination, desertion and conversion, thus optimize the network structure. The results of the calculating experimental demonstrate the effectiveness of the method.
出处
《华北水利水电学院学报》
1999年第4期44-48,共5页
North China Institute of Water Conservancy and Hydroelectric Power
