基于改进SGA的神经网络权系优化训练

IMPROVING SGA BASED-ON FINELY TRAINING PARAMETERS OF ANN

本文介绍了采用一种改进SGA来替代BP学习算法 ,对神经网络权系进行优化训练 ,仿真实验表明 ,SGA较BP收敛速度快 。

The parameters of ANN are finely trained with improving SGA in this paper, BP algorithm is replaced the improving SGA here. According to emulation Experiment. SGA is faster than BP in converging speed. It can refrain from the problem of partial minimum caused by BP.