Title :
Perceptron Multilayer Artificial Neural Network with Good Generalization
Author :
Yubo, Tian ; Yue, Dong ; Xiaoqiu, Zhang ; Renjie, Zhu
Author_Institution :
Jiangsu Univ. of Sci. & Technol., Zhenjiang
Abstract :
Feedforward perceptron multilayer (PML) ANNs with error back propagation learning method is often used in engineering design. Unfortunately, its generalization is often poor. In this paper, some reformative methods are proposed to improve its generalization. Based on the improved PML ANN, E-plane T-kind terminal matched load of rectangular waveguide is designed successfully. The result given by the ANN is agreeable with FDTD very well.
Keywords :
backpropagation; finite difference time-domain analysis; multilayer perceptrons; FDTD; error back propagation learning method; perceptron multilayer artificial neural network; rectangular waveguide; Artificial neural networks; Convergence; Finite difference methods; Learning systems; Least squares approximation; Microwave technology; Multi-layer neural network; Multilayer perceptrons; Neurons; Time domain analysis;
Conference_Titel :
Electromagnetics in Advanced Applications, 2007. ICEAA 2007. International Conference on
Conference_Location :
Torino
Print_ISBN :
978-1-4244-0767-5
Electronic_ISBN :
978-1-4244-0767-5
DOI :
10.1109/ICEAA.2007.4387431
