论文部分内容阅读
以分析输出OINN为基础、提出了分析前向神经网络隐节点数的线性方法,这种方法得出的隐节点数虽然只是充分条件,不是必要条件,但是得到的结果大大减少了网络不必要的隐节点数。
Based on the OINN output analysis, a linear method for analyzing the number of hidden nodes in the feedforward neural network is proposed. Although the number of hidden nodes obtained by this method is only sufficient and not necessary, the obtained result greatly reduces the unnecessary network Hidden nodes.