BP神经网络训练算法的分析与比较

BP神经网络被广泛应用于模式识别、信号处理和自动控制等领域，其广泛性是由于它能实现任何连续映射，但由于BP网络训练所固有的复杂性，目前尚没有任何一种完全的算法能适用于任何BP网络的训练。本文介绍了MATLAB神经网络工具箱中各种训练算法的特点及其函数的参数形式，并对它们的收敛速度和内存消耗情况进行了比较，说明了其各自适用的网络。

Analysis and Compare of BP Neural Network＇s Training Arithmetic

BP neural networks have been widely used in pattern recognition, signal processing, automatic control, etc. It owns universality, because it can make any continuous mapping come true, but, presently. there is not a compatible training arithmetic because of its complexity. This paper introduces the charac-teristics of training arithmetic in the MATLAB＇s Neural Network Toolbox and the parameters of their functions, then makes a compare of speed and memory among them, and explains which kinds of network they can be applied in.