Convergence of Offline Gradient Method with Inner-penalty for Multi-output Feedforward Neural Networks
Fengqi Zhou, Liu Er-gen, Xiaoyan Yu · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2013
The convergence of gradient method for the network training with one output unit has been considered by many authors [9][10][11][12][13][14].The convergence of the online and batch gradient algorithm with a penalty term for feedforward neural network Er gen Liu Yu Xiao Abstr act.In this paper, we study an offline gradient method with inner-penalty for training multi-output feedforward neural networks.The monotonicity of the error function and weight boundedness for the offline gradient with inner-penalty are presented, both weak and strong convergence results are proved, which will be very meaningful for theoretical research or applications on multi-output neural networks.