Integrated feature architecture selection

Author

Steppe, Jean M. ; Bauer, Kenneth W., Jr. ; Rogers, Steven K.

Author_Institution

Dept. of Oper. Sci., Air Force Inst. of Technol., Wright-Patterson AFB, OH, USA

Volume

7

Issue

4

fYear

1996

fDate

7/1/1996 12:00:00 AM

Firstpage

1007

Lastpage

1014

Abstract

In this paper, we present an integrated approach to feature and architecture selection for single hidden layer-feedforward neural networks trained via backpropagation. In our approach, we adopt a statistical model building perspective in which we analyze neural networks within a nonlinear regression framework. The algorithm presented in this paper employs a likelihood-ratio test statistic as a model selection criterion. This criterion is used in a sequential procedure aimed at selecting the best neural network given an initial architecture as determined by heuristic rules. Application results for an object recognition problem demonstrate the selection algorithm´s effectiveness in identifying reduced neural networks with equivalent prediction accuracy

Keywords

backpropagation; error statistics; feature extraction; feedforward neural nets; neural net architecture; object recognition; probability; statistical analysis; backpropagation; feature architecture selection; feedforward neural networks; heuristic rules; hidden layer; likelihood-ratio test; model selection criterion; nonlinear regression; object recognition; statistical model; Accuracy; Backpropagation algorithms; Buildings; Feedforward neural networks; Military computing; Multi-layer neural network; Neural networks; Object recognition; Statistical analysis; Testing;

fLanguage

English

Journal_Title

Neural Networks, IEEE Transactions on

Publisher

ieee

ISSN

1045-9227

Type

jour

DOI

10.1109/72.508942

Filename

508942