A Method of Improved BP Neural Algorithm Based on Simulated Annealing Algorithm

Author

Bai, Kai ; Xiong, Jing

Author_Institution

Sch. of Comput. Sci., Yangtze Univ., Jingzhou, China

fYear

2009

fDate

14-17 Oct. 2009

Firstpage

765

Lastpage

768

Abstract

This paper analyses the BP algorithm in detail, including the number of hidden layer, the amount of neural node and training algorithm. In order to improve the training speed, this paper adopts the automatic and adaptive step to perfect the BP algorithm. In addition, because the traditional BP neural network is easy to trap into local minimum, this paper makes use of the characteristic of simulated annealing algorithm and let it unite with BP algorithm. Because the simulated annealing algorithm can get optimal approximation by searching local, it can help BP algorithm not to trap into local minimum.

Keywords

backpropagation; neural nets; simulated annealing; BP neural network; neural node; neural training algorithm; simulated annealing; Analytical models; Approximation algorithms; Artificial neural networks; Biological neural networks; Biological system modeling; Computational modeling; Computer science; Computer simulation; Genetics; Simulated annealing; BP Neural Algorithm; neural network; simulated annealing algorithm;

fLanguage

English

Publisher

ieee

Conference_Titel

Genetic and Evolutionary Computing, 2009. WGEC '09. 3rd International Conference on

Conference_Location

Guilin

Print_ISBN

978-0-7695-3899-0

Type

conf

DOI

10.1109/WGEC.2009.39

Filename

5402822