Design and evaluation of neural classifiers

Author

Hintz-Madsen, Mads ; Pedersen, Morten With ; Hansen, Lars Kai ; Larsen, Jan

Author_Institution

Inst. of Math. Modeling, Tech. Univ., Lyngby, Denmark

fYear

1996

fDate

4-6 Sep 1996

Firstpage

223

Lastpage

232

Abstract

In this paper we propose a method for the design of feedforward neural classifiers based on regularization and adaptive architectures. Using a penalized maximum likelihood scheme we derive a modified form of the entropy error measure and an algebraic estimate of the test error. In conjunction with optimal brain damage pruning the test error estimate is used to optimize the network architecture. The scheme is evaluated on an artificial and a real world problem

Keywords

adaptive systems; entropy; error analysis; feedforward neural nets; maximum likelihood estimation; neural net architecture; optimisation; pattern classification; adaptive architectures; algebraic error estimate; entropy error measure; feedforward neural networks; glass classification; neural classifiers; optimal brain damage pruning; pattern classification; penalized maximum likelihood estimation; Artificial neural networks; Biological neural networks; Computer architecture; Feedforward systems; Frequency estimation; Maximum likelihood estimation; Optimization methods; Pattern recognition; Probability distribution; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks for Signal Processing [1996] VI. Proceedings of the 1996 IEEE Signal Processing Society Workshop

Conference_Location

Kyoto

ISSN

1089-3555

Print_ISBN

0-7803-3550-3

Type

conf

DOI

10.1109/NNSP.1996.548352

Filename

548352