A New Conjugate Gradient Method for Learning Fuzzy Neural Networks

Authors : Hisham MOHAMMED, Khalil K ABBO

Pages : 57-69

View : 7 | Download : 2

Publication Date : 2021-03-25

Article Type : Research Paper

Abstract :In this paper, we suggest a conjugate gradient method, which belongs to the op-timization methods for learning a fuzzy neural network model that is based on Takagi Sugeno. Where we developed a new algorithm based on the Polak–Ribière–Polak insert ignore into journalissuearticles values(PRP); method, The technique developed is converging by assum-ing a certain hypothesis. The numerical results indicate the efficacy of the method developed for classifying data as shown in the table as the new method was supe-rior to the Polak–Ribière–Polak insert ignore into journalissuearticles values(PRP); and Liu-Storey insert ignore into journalissuearticles values(LS); methods in average training time, Average training accuracy, Average test accuracy, Average train-ing MSE, and Average test MSE. As for the figures, we showed the superiority of the new algorithm in The average training accuracy and The average training error Compared to Polak–Ribière–Polak insert ignore into journalissuearticles values(PRP); and Liu-Storey insert ignore into journalissuearticles values(LS); methods, in 100 No. of training iteration.
Keywords : algorithm, classification, fuzzy neural networks, techniques, optimization