Global optimization for neural network training

Author

Shang, Yi ; Wah, Benjamin W.

Author_Institution

Illinois Univ., Champaign, IL, USA

Volume

29

Issue

3

fYear

1996

fDate

3/1/1996 12:00:00 AM

Firstpage

45

Lastpage

54

Abstract

We propose a novel global minimization method, called NOVEL (Nonlinear Optimization via External Lead), and demonstrate its superior performance on neural network learning problems. The goal is improved learning of application problems that achieves either smaller networks or less error prone networks of the same size. This training method combines global and local searches to find a good local minimum. In benchmark comparisons against the best global optimization algorithms, it demonstrates superior performance improvement

Keywords

learning (artificial intelligence); minimisation; neural nets; nonlinear programming; search problems; NOVEL; Nonlinear Optimization via External Lead; application problems; benchmark comparison; global minimization method; local minimum; local searches; neural network learning problems; neural network training; Feedforward neural networks; Feedforward systems; Heuristic algorithms; Minimization methods; Neural networks; Optimization methods; Search methods; Supervised learning; Testing; Topology;

fLanguage

English

Journal_Title

Computer

Publisher

ieee

ISSN

0018-9162

Type

jour

DOI

10.1109/2.485892

Filename

485892