Linear discriminant analysis for two classes via recursive neural network reduction of the class separation

Mayer Aladjem

doi:10.1007/bfb0033302

Linear discriminant analysis for two classes via recursive neural network reduction of the class separation

Mayer Aladjem

Department of Electrical & Computer Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

2 Scopus citations

Abstract

A method for the linear discrimination of two classes is presented. It maximizes the Patrick-Fisher (PF) distance between the projected class-conditional densities. Since the PF distance is a highly nonlinear function, we propose a method, which searches for the directions corresponding to several large local maxima of the PF distance. Its novelty lies in a neural network transformation of the data along a found direction into data with deflated maxima of the PF distance and iteration to obtain the next direction. A simulation study indicates that the method has the potential to find the global maximum of the PF distance.

Original language	English
Title of host publication	Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings
Editors	Adnan Amin, Dov Dori, Pavel Pudil, Herbert Freeman
Publisher	Springer Verlag
Pages	775-784
Number of pages	10
ISBN (Print)	3540648585, 9783540648581
DOIs	https://doi.org/10.1007/bfb0033302
State	Published - 1 Jan 1998
Event	7th Joint IAPR International Workshop on Structural and Syntactic Pattern Recognition, SSPR 1998 and 2nd International Workshop on Statistical Techniques in Pattern Recognition, SPR 1998 - Sydney, Australia Duration: 11 Aug 1998 → 13 Aug 1998

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	1451
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	7th Joint IAPR International Workshop on Structural and Syntactic Pattern Recognition, SSPR 1998 and 2nd International Workshop on Statistical Techniques in Pattern Recognition, SPR 1998
Country/Territory	Australia
City	Sydney
Period	11/08/98 → 13/08/98

Keywords

Auto-associative network
Discriminant analysis
Neural networks for classification
Projection pursuit
Statistical pattern recognition

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/bfb0033302

Cite this

Aladjem, M. (1998). Linear discriminant analysis for two classes via recursive neural network reduction of the class separation. In A. Amin, D. Dori, P. Pudil, & H. Freeman (Eds.), Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings (pp. 775-784). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1451). Springer Verlag. https://doi.org/10.1007/bfb0033302

Aladjem, Mayer. / Linear discriminant analysis for two classes via recursive neural network reduction of the class separation. Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings. editor / Adnan Amin ; Dov Dori ; Pavel Pudil ; Herbert Freeman. Springer Verlag, 1998. pp. 775-784 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{f96360a910cc416abc8679d8bbdd1e0a,

title = "Linear discriminant analysis for two classes via recursive neural network reduction of the class separation",

abstract = "A method for the linear discrimination of two classes is presented. It maximizes the Patrick-Fisher (PF) distance between the projected class-conditional densities. Since the PF distance is a highly nonlinear function, we propose a method, which searches for the directions corresponding to several large local maxima of the PF distance. Its novelty lies in a neural network transformation of the data along a found direction into data with deflated maxima of the PF distance and iteration to obtain the next direction. A simulation study indicates that the method has the potential to find the global maximum of the PF distance.",

keywords = "Auto-associative network, Discriminant analysis, Neural networks for classification, Projection pursuit, Statistical pattern recognition",

author = "Mayer Aladjem",

note = "Publisher Copyright: {\textcopyright} Springer-Verlag Berlin Heidelberg 1998.; 7th Joint IAPR International Workshop on Structural and Syntactic Pattern Recognition, SSPR 1998 and 2nd International Workshop on Statistical Techniques in Pattern Recognition, SPR 1998 ; Conference date: 11-08-1998 Through 13-08-1998",

year = "1998",

month = jan,

day = "1",

doi = "10.1007/bfb0033302",

language = "English",

isbn = "3540648585",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "775--784",

editor = "Adnan Amin and Dov Dori and Pavel Pudil and Herbert Freeman",

booktitle = "Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings",

address = "Germany",

}

Aladjem, M 1998, Linear discriminant analysis for two classes via recursive neural network reduction of the class separation. in A Amin, D Dori, P Pudil & H Freeman (eds), Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1451, Springer Verlag, pp. 775-784, 7th Joint IAPR International Workshop on Structural and Syntactic Pattern Recognition, SSPR 1998 and 2nd International Workshop on Statistical Techniques in Pattern Recognition, SPR 1998, Sydney, Australia, 11/08/98. https://doi.org/10.1007/bfb0033302

Linear discriminant analysis for two classes via recursive neural network reduction of the class separation. / Aladjem, Mayer.
Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings. ed. / Adnan Amin; Dov Dori; Pavel Pudil; Herbert Freeman. Springer Verlag, 1998. p. 775-784 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1451).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Linear discriminant analysis for two classes via recursive neural network reduction of the class separation

AU - Aladjem, Mayer

N1 - Publisher Copyright: © Springer-Verlag Berlin Heidelberg 1998.

PY - 1998/1/1

Y1 - 1998/1/1

N2 - A method for the linear discrimination of two classes is presented. It maximizes the Patrick-Fisher (PF) distance between the projected class-conditional densities. Since the PF distance is a highly nonlinear function, we propose a method, which searches for the directions corresponding to several large local maxima of the PF distance. Its novelty lies in a neural network transformation of the data along a found direction into data with deflated maxima of the PF distance and iteration to obtain the next direction. A simulation study indicates that the method has the potential to find the global maximum of the PF distance.

AB - A method for the linear discrimination of two classes is presented. It maximizes the Patrick-Fisher (PF) distance between the projected class-conditional densities. Since the PF distance is a highly nonlinear function, we propose a method, which searches for the directions corresponding to several large local maxima of the PF distance. Its novelty lies in a neural network transformation of the data along a found direction into data with deflated maxima of the PF distance and iteration to obtain the next direction. A simulation study indicates that the method has the potential to find the global maximum of the PF distance.

KW - Auto-associative network

KW - Discriminant analysis

KW - Neural networks for classification

KW - Projection pursuit

KW - Statistical pattern recognition

UR - http://www.scopus.com/inward/record.url?scp=33749939659&partnerID=8YFLogxK

U2 - 10.1007/bfb0033302

DO - 10.1007/bfb0033302

M3 - Conference contribution

AN - SCOPUS:33749939659

SN - 3540648585

SN - 9783540648581

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 775

EP - 784

BT - Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings

A2 - Amin, Adnan

A2 - Dori, Dov

A2 - Pudil, Pavel

A2 - Freeman, Herbert

PB - Springer Verlag

T2 - 7th Joint IAPR International Workshop on Structural and Syntactic Pattern Recognition, SSPR 1998 and 2nd International Workshop on Statistical Techniques in Pattern Recognition, SPR 1998

Y2 - 11 August 1998 through 13 August 1998

ER -

Aladjem M. Linear discriminant analysis for two classes via recursive neural network reduction of the class separation. In Amin A, Dori D, Pudil P, Freeman H, editors, Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings. Springer Verlag. 1998. p. 775-784. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/bfb0033302

Linear discriminant analysis for two classes via recursive neural network reduction of the class separation

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this