论文部分内容阅读
考虑三层前馈神经网络隐结点学习问题．在分析同类与不同类训练样本在隐层输出上体现的差异的基础上，提出了一种在权值学习过程中动态地用除网络隐结点数的学习算法．数值结果表明本文算法是可行的．
Consider the hidden layer learning problem of three layers feedforward neural network. Based on the analysis of the differences between the hidden samples and the hidden samples output by the same or different training samples, a learning algorithm that dynamically uses hidden nodes in the weighted learning process is proposed. Numerical results show that the proposed algorithm is feasible.