Title :
A New NN Fuzzy BP Learning Approach
Author :
Weng, Pin-Hsuan ; Chen, Yu-Ju ; Chang, Chuo-Yean ; Hwang, Rey-Chue
Author_Institution :
I-Shou Univ., Kaohsiung
Abstract :
In this paper, a new approach of fuzzy back-propagation learning method of neural network is proposed. The proposed algorithm not only improves the speed of learning capability, but also increases the accuracy of signal prediction. Several experiments were conducted by this new learning rule. For comparison, parallel simulations were conducted with the same data using a standard back-propagation (BP) algorithm, with constant learning rates (alpha=0.1~0.9) and momentums (xi=0.1~0.9).
Keywords :
backpropagation; fuzzy neural nets; fuzzy back-propagation learning method; neural network; signal prediction; Acceleration; Accuracy; Artificial neural networks; Convergence; Electronic mail; Fuzzy neural networks; Fuzzy systems; Information management; Learning systems; Neural networks;
Conference_Titel :
Innovative Computing, Information and Control, 2007. ICICIC '07. Second International Conference on
Conference_Location :
Kumamoto
Print_ISBN :
0-7695-2882-1
DOI :
10.1109/ICICIC.2007.55
