摘要： 提出一种改进的BP算法,并对改进算法中的参数选择进行了讨论。改进后的算法用于XOR问题的学习及函数逼近问题。结果表明,改进后的BP算法可显著地提高网络的学习速度和逼近精度。

关键词: 神经网络, BP算法, 相对熵, 误差函数

Abstract: An improved BP algorithm is proposed adopting cross entropy function as error function.Moreover,parameter selection in the improved algorithm is discussed.The improved algorithm is applied to the learning of the XOR question and function approximation question.The simulation results show that this algorithm has the advantages in fast convergent speed and high approximation accuracy of the network.

中图分类号: 

• TP183