GXqLTg8yWwp6OMZqUPSDDlzwHDLfYH9IfioIQIZBMFkqyRTSLd1DjL2QfgSm
Current position: Home >> Scientific Research >> Paper Publications

带惩罚项与随机输入的BP神经网络在线梯度学习算法的收敛性

Release Time:2019-03-10  Hits:

Indexed by: Journal Article

Date of Publication: 2007-08-15

Journal: 数学研究与评论

Included Journals: CSCD

Volume: 27

Issue: 3

Page Number: 643-653

ISSN: 1000-341X

Key Words: BP神经网络;在线梯度法;收敛性;惩罚项;随机输入.

Abstract: 本文对三层BP神经网络中带有惩罚项的在线梯度学习算法的收敛性问题进行了研究.在网络训练每一轮开始执行之前,对训练样本随机进行重排,以使网络学习更容易跳出局部极小.文中给出了误差函数的单调性定理以及该算法的弱收敛和强收敛性定理.

Prev One:Uniqueness of linear combinations of ridge functions

Next One:Approximation to a compact set of functions by feedforward neural networks