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