Title :
Incorporating additional hint neurons in recurrent neural networks to improve convergence
Author :
Zhao, Songhe ; Dillon, T.S.
Author_Institution :
Expert & Intelligent Syst. Lab., La Trobe Univ., Bundoora, Vic., Australia
Abstract :
The approach to determining a neural network involves using only the training patterns or examples. In this paper, the authors propose an approach to incorporate additional hint neurons into the proposed recurrent neural network. Experiments have been conducted on the oscillation problem. The results show that with the help of the hint function, the network learns to model the oscillator with greater ease
Keywords :
convergence; learning (artificial intelligence); multilayer perceptrons; recurrent neural nets; convergence; hint neurons; oscillation problem; recurrent neural networks; training patterns; Convergence; Feedforward systems; Intelligent networks; Intelligent systems; Laboratories; Multi-layer neural network; Neural networks; Neurons; Nonhomogeneous media; Recurrent neural networks;
Conference_Titel :
Neural Networks, 1995. Proceedings., IEEE International Conference on
Conference_Location :
Perth, WA
Print_ISBN :
0-7803-2768-3
DOI :
10.1109/ICNN.1995.487350