Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
A Comparative Study of Neural Network Based Feature Extraction Paradigms Boaz Lerner*, Hugo Guterman#, Mayer Aladjem#, and Its'hak Dinstein#
 

Summary: A Comparative Study of Neural Network Based Feature Extraction Paradigms
Boaz Lerner*, Hugo Guterman#, Mayer Aladjem#, and Its'hak Dinstein#
*University of Cambridge Computer Laboratory, New Museums Site, Cambridge CB2 3QG, UK
#Department of Electrical and Computer Engineering, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel
Published in Pattern Recognition Letters, vol. 20(1), pp. 7-14, 1999.
Abstract
The projection maps and derived classification accuracies of a neural network (NN) implementation
of Sammon's mapping, an auto-associative NN (AANN) and a multilayer perceptron (MLP) feature
extractor are compared with those of the conventional principal component analysis (PCA). Tested on
five real-world databases, the MLP provides the highest classification accuracy at the cost of deforming
the data structure, whereas the linear models preserve the structure but usually with inferior accuracy.
Keywords: Auto-associative neural network; Classification; Data projection; Feature extraction; Multilayer
perceptron; Principal components; Sammon's mapping;
1. Introduction
The process of mapping original features (measurements) into fewer, more effective features is
termed feature extraction. In each of the existing feature extraction methods (Fukunaga, 1990, Ch. 9-
10), a mapping f transforms a d-dimensional pattern X to an m-dimensional pattern Y (m such that a criterion J is optimised. Examples of such a criterion are the mean square error (used for
example in PCA) and the inter-pattern distance error used in Sammon's mapping. The mapping f is
determined from among all the transformations g as one that satisfies,

  

Source: Aladjem, Mayer - Department of Electrical and Computer Engineering, Ben-Gurion University

 

Collections: Computer Technologies and Information Sciences