摘要
针对前向神经网络现有BP学习算法的不足,结合非线性最优化方法,提出一种基于拟牛顿法的神经元网络学习算法。该算法有效地改进了神经元网络的学习收敛速度,取得了比常规BP算法更好的收敛性能和学习速度。
Since the convergence speed of the BP learning algorithm is slow, the data stability appears to be very poor. A new algorithm based on quasi -Newton method is proposed. The new algorithm, compared to the BP algorithm, has the fast learning rate and good convergence properties.
出处
《控制与决策》
EI
CSCD
北大核心
1997年第4期357-360,共4页
Control and Decision
关键词
神经网络
拟牛顿法
反向传播算法
neural networks, quasi -Newton method, back -propagation algorithm
