中文说明：RNN网络连接权值的学习修正使用BP算法，由于网络隐含层结点中的作用函数采用高斯函数，从而具有局部逼近能力．此为该网络之所以学习速度快的原因．此外．由于GRNN中人为调节参数很少．只有一个阈值．网络的学习全部依耐数据样本，这个特点决定网络得以最大可能地避免人为主观假定对预测结果的影响。
English Description：
Learning correction RNN network connection weights using BP algorithm, due to the network nodes in the hidden layer function takes the role of a Gaussian function, which has a capacity of local approximation. The reason for the network to learn the reasons for speed. In addition. Because few adjustable parameters GRNN artificially. Only one threshold value. All learning network according to resistance data samples, this feature determines the network to the maximum extent possible to avoid human subjective assumptions affect prediction results.