Convergence analysis of a discrete-time recurrent neural network to perform quadratic real optimization with bound constraints

Author

Pérez-Ilzarbe, María José

Author_Institution

Dept. de Autom. y Comput., Univ. Publica de Navarra, Spain

Volume

9

Issue

6

fYear

1998

fDate

11/1/1998 12:00:00 AM

Firstpage

1344

Lastpage

1351

Abstract

Presents a model of a discrete-time recurrent neural network designed to perform quadratic real optimization with bound constraints. The network iteratively improves the estimate of the solution, always maintaining it inside of the feasible region. Several neuron updating rules which assure global convergence of the net to the desired minimum have been obtained. Some of them also assure exponential convergence and maximize a lower bound for the convergence degree. Simulation results are presented to show the net performance

Keywords

convergence of numerical methods; discrete time systems; gradient methods; optimisation; recurrent neural nets; bound constraints; convergence analysis; convergence degree; discrete-time recurrent neural network; exponential convergence; global convergence; neuron updating rules; quadratic real optimization; Constraint optimization; Convergence of numerical methods; Design optimization; Gradient methods; Lagrangian functions; Neural networks; Neurons; Optimization methods; Performance analysis; Recurrent neural networks;

fLanguage

English

Journal_Title

Neural Networks, IEEE Transactions on

Publisher

ieee

ISSN

1045-9227

Type

jour

DOI

10.1109/72.728385

Filename

728385