BP网络局部极小值的修正算法

Implementation of Modification Algorithm for BP Network Local Minima

在人工神经网络中,反向传播法占据了非常重要的地位。近几十年许多研究人员对其做了深入的研究,提出了很多改进的方法,主要目标是为了加快训练速度,避免陷入局部极小值和改善其他能力。本文主要讨论在MATLAB下使用附加动量法使网络避免陷入局部极小值问题。

BP Network plays a very important role in artificial neural networks. In recent several decades, many researchers have carried out deepresearch on it, and put many improved methods, aiming at accelerating training process, and avoiding local minima and improving other abilities. This article discusses how to avoid local minima using additional momentum method with the help of MATLAB.