Abstract—A major problem of neural network in real-time applications is their long training time. We present a modification of the neural network (NN) for reduction of training time. Unlike traditional training time reduction algorithms, we propose a new Adaptive Fuzzy technique to create ensembles of neural network using multiple projections of the same data obtained from different NNs. The purpose of this paper is to demonstrate the optimization of training that occurs with the application of fuzzy controller theory to neural network. A fuzzy system is employed to control the learning parameters of a neural network to reduce the possibility of overshooting during the learning process. Hence, the learning time can be shortened. This paper compares the training efficiency and accuracy between a NN and a fuzzy controlled neural network, when they are required to carry out the same assignment. We justify the suitability of the proposed method by some experiments in soccer robot trajectory generation tasks; the resulting fuzzy controlled neural network indicates a significant reduction in the training time by 30%.
Index Terms—Neural Network, Adaptive Fuzzy Logic Controller, Backpropagation Learning Algorithm, Mobile Robot, Trajectory Generation.
H. R. Kanan is with the Department of Electrical Engineering, Bu-Ali Sina University, Hamedan, Iran (Tel.+98-811 8257409; fax: +98-8118257400; 65175-4161; email: firstname.lastname@example.org).
M. Y. A. Khanian is with the Electrical, Computer and IT Engineering Department, Qazvin Branch, Islamic Azad University, Qazvin, Iran.
Cite: Hamidreza Rashidy Kanan and Mahdi Yousefi Azar Khanian, "Reduction of Neural Network Training Time Using an Adaptive Fuzzy Approach in Real Time Applications," International Journal of Information and Electronics Engineering, vol. 2, no. 3, pp. 470-474, 2012.