An Improved BP Algorithm Based on the Nonlinear Least Square Method and Application in Congestion Control

Author

Cai, Haibin ; Cao, Qiying

Author_Institution

Coll. of Inf., Donghua Univ., Shanghai

Volume

1

fYear

0

fDate

0-0 0

Firstpage

2830

Lastpage

2834

Abstract

The standard back-propagation (BP) algorithm converges slowly and is ease to trip into local minimum, which is the main reason why it cannot be used widely in practical applications. Therefore, a new BP algorithm based on the nonlinear least square method was put forward in this paper. The convergence stability was enhanced by adding the positive matrix in Hessian matrix. The results have proved that the new algorithm can converges very fast and avoid tripping into local minimum. The improved BP algorithm is used in congestion control of the computer network in order to solve the congestion of network. The data of simulation experiment demonstrate the proposed control scheme is scalable and efficient, and has good network performance

Keywords

Hessian matrices; backpropagation; computer networks; convergence; least squares approximations; neurocontrollers; stability; telecommunication congestion control; Hessian matrix; backpropagation algorithm; computer network; congestion control; convergence stability; learning; neural network; nonlinear least square method; positive matrix; Algorithm design and analysis; Application software; Cities and towns; Computer science; Educational institutions; Error correction; Function approximation; Least squares methods; Neural networks; Neurons; Neural network; back-propagation algorithm; congestion control; learning rate; nonlinear least square method;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Control and Automation, 2006. WCICA 2006. The Sixth World Congress on

Conference_Location

Dalian

Print_ISBN

1-4244-0332-4

Type

conf

DOI

10.1109/WCICA.2006.1712881

Filename

1712881