Improving discriminant neural network (DNN) design by the use of principal component analysis

Author

Li, Qi ; Tufts, Donald W.

Author_Institution

Dept. of Electr. Eng., Rhode Island Univ., Kingston, RI, USA

Volume

5

fYear

1995

fDate

9-12 May 1995

Firstpage

3375

Abstract

Investigations of the design of a discriminant neural network (DNN) have shown the advantages of sequential design of hidden nodes and pruning of the training data for improved classification and fast training time. The performance can be further improved by adding the capability of a nonlinear, principal component discriminant node. This type of hidden node is useful for separating classes which have common mean vectors and are overlapped on one other

Keywords

learning (artificial intelligence); neural net architecture; classification; discriminant neural network design; fast training time; hidden node; hidden nodes; mean vectors; neural network architecture; nonlinear principal component discriminant node; performance; principal component analysis; sequential design; training data pruning; Analytical models; Backpropagation; Covariance matrix; Design methodology; Eigenvalues and eigenfunctions; Linear discriminant analysis; Neural networks; Principal component analysis; Training data; Vectors;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1995. ICASSP-95., 1995 International Conference on

Conference_Location

Detroit, MI

ISSN

1520-6149

Print_ISBN

0-7803-2431-5

Type

conf

DOI

10.1109/ICASSP.1995.479709

Filename

479709