变学习率学习法则是近年来提出并被验证的适用于对抗系统的学习法则。
WOLF (Win or Learn Fast) is a newly proposed learning principle which is proven to be suitable for adversarial system.
就训练次数与精确度而言,它明显优于共轭梯度法及变学习率的BP算法,适用于系统辨识。
Concerned with the training process and accuracy, the LM algorithm is superior to conjugate gradient algorithm and a variable learning rate back propagation (BP) algorithm.
通过离线的迭代算法生成高精度的样本点来训练神经网络,使用动量法、变学习率法和共轭梯度法提高BP网络的收敛速度。
Methods based on BP neural network and RBF neural network were studied to solve inverse kinematics. The training samples were obtained through off-line numerical method with high precision.
应用推荐