Asia-Pacific Youth Conference on Communication Technology (APYCCT 2010 E-BOOK)

Kunming,China,8.7-8.8,2010

ISBN: 978-1-935068-20-4 Scientific Research Publishing, USA

E-Book 934pp Pub. Date: August 2010

Category: Computer Science & Communications

Price: $120

Title: A General Prediction Method Based on GA-BP Network
Source: Asia-Pacific Youth Conference on Communication Technology (APYCCT 2010 E-BOOK)(Part 3 Computer Technology and Application) (pp 219-222)
Author(s): Hai-ru Guo, School of computer, Xiaogan University, Xiaogan, China
Xue-mei Cui, School of life science technology, Xiaogan University, Xiaogan, China
Abstract: A new general method for prediction was set up on the study of annual enrollment scale in Xiaogan University. LM-BP network learns well but predicts badly. The main reason for it is that LM-BP network is sensitive to its initial weights and thresholds. Thus genetic algorithm (GA) was adopted to optimize the initial weights and thresholds of LM-BP network, and the general method for prediction based on GA-BP network was proposed. Compared with the method that the initial weights and thresholds were selected randomly, the new general method has the following advantages: substantial amount of time can be reduced, and it has high measurement precision. After the initial weights and thresholds were optimized by GA, the generalization ability of neural network can be greatly improved. By using the new method, the prediction for enrollment scale in Xiaogan University has certain reference value.
Free SCIRP Newsletters
Copyright © 2006-2024 Scientific Research Publishing Inc. All Rights Reserved.
Top