论文部分内容阅读
当训练样本线性可分时,本文证明前馈神经网络的在线BP算法是有限次收敛的．
When the training samples are linearly separable, this paper proves that the online BP algorithm of feedforward neural network converges to a finite degree.