几种快速BP算法的比较研究  

Study on Comparison of Fast BP Algorithms

作　　者：吕俊[1] 张兴华[1] 

机构地区：[1]南京工业大学,江苏南京210009

出　　处：《现代电子技术》2003年第24期96-99,共4页Modern Electronics Technique

摘　　要：对几种快速 BP算法的特点及性能作了归纳和对比 ,并对一个非线性函数逼近实例进行了仿真研究。结果表明对于中等规模的前向神经网络来说 ,L evenberg_ Marquardt算法收敛速度最快 ,而且学习性能最好。In this paper, the characteristic and performance of various fast BP algorithms are generalized and contrasted through study on simulation of nonlinear function approximation experiment. It is demonstrated LevenbergMarquardt algorithm′s convergence speed is the quickest and its performance is the most excellent for moderate scale feedforward neural network.

关 键 词：神经网络 权值学习 快速BP算法 Levenberg-Marqardt算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]