一种BP神经网络算法的改进模型

首先对BP网络的结构和算法进行了分析,针对BP网络收敛速度慢,容易陷入局部极小等问题,提出了一种改进的BP网络模型,并对该模型算法进行了改进,通过激活函数的选择,网络的初始化,学习率的调整和训练样本数据的处理等方法,可实现加快网络的收敛速度,并且较好的解决局部最优问题.

An Improved Model of BP Neural Network Algorithm

Based on analyzing structure and Algorithm of BP network, as for these problems of slow convergence and easily getting into the local minimum in the study algorithm of BP net, this paper designs one improved model of BP neural network By choosing suitable activated function, taking its sensibility into accounts, and standing out researching the net's initializing matter etc, the faster convergence speed could be implemented and the problem of local optimization be solved better.