神经网络中引入数值积分寻优法  

Introduction of Numerical Integral Optimization In Neural Network

作　　者：王学峰[1] 张峰[1] 

机构地区：[1]石河子大学师院/兵团教院,新疆石河子832003

出　　处：《兵团教育学院学报》2003年第1期38-40,共3页Journal of Bingtuan Education Institute

摘　　要：本文提出了一种新的寻优方法———数值积分寻优法 ,将此方法应用于神经网络的学习算法中 ,构造了两个神经网络 ,它们与BP网络有相同功能 ,且不出现BP网络的局部极小问题 。This paper introduces that a new optimal method--numerical integral optimization is applied to neural network and two neural networks are constructed. It also explains that they have the same function as BP network but can overcome some defects of it and their convergent speed is faster than it.

关 键 词：神经网络 数值积分寻优法 算法 IBP网 

分 类 号：O241.4[理学—计算数学]