BP网络的主控式误差调整学习算法  

Active Control BP Learning Algorithm for Feed-forward Neural Network

作　　者：孙巧萍[1] 

机构地区：[1]台州学院计算机系,浙江临海317000

出　　处：《台州学院学报》2002年第6期22-23,27,共3页Journal of Taizhou University

摘　　要：提出一种BP神经网络的主控式误差调整学习算法,该算法根据网络输出端的误差变化趋势主动控制输出层的误差调整,进而由调整误差的主动调节控制网络权值和阈值的学习。计算机模拟结果表明,这种学习方法具有能稳定收敛、收敛速度快、适用范围宽、泛化能力强等特点,从而在一定程度上克服了传统BP算法的局限性,实现较有效的学习。According to the error - changing tendency of the output layer, control BP algorithm for adjusting error(ACBP) controls the error adjustment of the output layer actively, and the active error - adjuster further modulates the learn-ing of network weights. The experiments show that ACBP learning algorithm is characterized by a steady and rapid conver-gence, a wide range of application, and a strong generalized capacity, which, to a certain degree, overcomes the disadvanta-ges of the traditional BP algorithm and offers the relatively effective learning.

关 键 词：BP神经网络 误差调整 收敛速度 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]