Fast learning fully complex-valued classifiers for real-valued classification problems

R. Savitha, S. Suresh, N. Sundararajan, Hyong Joong Kim

Research output: Chapter in Book/Report/Conference proceedingConference contribution

27 Citations (Scopus)

Abstract

In this paper, we present two fast learning neural network classifiers with a single hidden layer: the 'Phase Encoded Complex-valued Extreme Learning Machine (PE-CELM)' and the 'Bilinear Branch-cut Complex-valued Extreme Learning Machine (BB-CELM)'. The proposed classifiers use the phase encoded transformation and the bilinear transformation with a branch-cut at 2π as the activation functions in the input layer to map the real-valued features to the complex domain. The neurons in the hidden layer employ the fully complex-valued activation function of the type of a hyperbolic secant function. The parameters of the hidden layer are chosen randomly and the output weights are estimated as the minimum norm least square solution to a set of linear equations. The classification ability of these classifiers are evaluated using a set of benchmark data sets from the UCI machine learning repository. Results highlight the superior classification ability of these classifiers with least computational effort.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages602-609
Number of pages8
Volume6675 LNCS
EditionPART 1
DOIs
Publication statusPublished - 2011 Jun 6
Event8th International Symposium on Neural Networks, ISNN 2011 - Guilin, China
Duration: 2011 May 292011 Jun 1

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 1
Volume6675 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other8th International Symposium on Neural Networks, ISNN 2011
CountryChina
CityGuilin
Period11/5/2911/6/1

Fingerprint

Classification Problems
Classifiers
Classifier
Extreme Learning Machine
Learning systems
Activation Function
Secant function
Branch
Chemical activation
Hyperbolic secant
Hyperbolic functions
Hyperbolic function
Least-squares Solution
Phase Transformation
Linear equations
Repository
Neurons
Neuron
Linear equation
Machine Learning

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Savitha, R., Suresh, S., Sundararajan, N., & Kim, H. J. (2011). Fast learning fully complex-valued classifiers for real-valued classification problems. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 1 ed., Vol. 6675 LNCS, pp. 602-609). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6675 LNCS, No. PART 1). https://doi.org/10.1007/978-3-642-21105-8_70

Fast learning fully complex-valued classifiers for real-valued classification problems. / Savitha, R.; Suresh, S.; Sundararajan, N.; Kim, Hyong Joong.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 6675 LNCS PART 1. ed. 2011. p. 602-609 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6675 LNCS, No. PART 1).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Savitha, R, Suresh, S, Sundararajan, N & Kim, HJ 2011, Fast learning fully complex-valued classifiers for real-valued classification problems. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 edn, vol. 6675 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 1, vol. 6675 LNCS, pp. 602-609, 8th International Symposium on Neural Networks, ISNN 2011, Guilin, China, 11/5/29. https://doi.org/10.1007/978-3-642-21105-8_70
Savitha R, Suresh S, Sundararajan N, Kim HJ. Fast learning fully complex-valued classifiers for real-valued classification problems. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 ed. Vol. 6675 LNCS. 2011. p. 602-609. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1). https://doi.org/10.1007/978-3-642-21105-8_70
Savitha, R. ; Suresh, S. ; Sundararajan, N. ; Kim, Hyong Joong. / Fast learning fully complex-valued classifiers for real-valued classification problems. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 6675 LNCS PART 1. ed. 2011. pp. 602-609 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).
@inproceedings{78b60f1d8d95498aa083f178ab17b657,
title = "Fast learning fully complex-valued classifiers for real-valued classification problems",
abstract = "In this paper, we present two fast learning neural network classifiers with a single hidden layer: the 'Phase Encoded Complex-valued Extreme Learning Machine (PE-CELM)' and the 'Bilinear Branch-cut Complex-valued Extreme Learning Machine (BB-CELM)'. The proposed classifiers use the phase encoded transformation and the bilinear transformation with a branch-cut at 2π as the activation functions in the input layer to map the real-valued features to the complex domain. The neurons in the hidden layer employ the fully complex-valued activation function of the type of a hyperbolic secant function. The parameters of the hidden layer are chosen randomly and the output weights are estimated as the minimum norm least square solution to a set of linear equations. The classification ability of these classifiers are evaluated using a set of benchmark data sets from the UCI machine learning repository. Results highlight the superior classification ability of these classifiers with least computational effort.",
author = "R. Savitha and S. Suresh and N. Sundararajan and Kim, {Hyong Joong}",
year = "2011",
month = "6",
day = "6",
doi = "10.1007/978-3-642-21105-8_70",
language = "English",
isbn = "9783642211041",
volume = "6675 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 1",
pages = "602--609",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
edition = "PART 1",

}

TY - GEN

T1 - Fast learning fully complex-valued classifiers for real-valued classification problems

AU - Savitha, R.

AU - Suresh, S.

AU - Sundararajan, N.

AU - Kim, Hyong Joong

PY - 2011/6/6

Y1 - 2011/6/6

N2 - In this paper, we present two fast learning neural network classifiers with a single hidden layer: the 'Phase Encoded Complex-valued Extreme Learning Machine (PE-CELM)' and the 'Bilinear Branch-cut Complex-valued Extreme Learning Machine (BB-CELM)'. The proposed classifiers use the phase encoded transformation and the bilinear transformation with a branch-cut at 2π as the activation functions in the input layer to map the real-valued features to the complex domain. The neurons in the hidden layer employ the fully complex-valued activation function of the type of a hyperbolic secant function. The parameters of the hidden layer are chosen randomly and the output weights are estimated as the minimum norm least square solution to a set of linear equations. The classification ability of these classifiers are evaluated using a set of benchmark data sets from the UCI machine learning repository. Results highlight the superior classification ability of these classifiers with least computational effort.

AB - In this paper, we present two fast learning neural network classifiers with a single hidden layer: the 'Phase Encoded Complex-valued Extreme Learning Machine (PE-CELM)' and the 'Bilinear Branch-cut Complex-valued Extreme Learning Machine (BB-CELM)'. The proposed classifiers use the phase encoded transformation and the bilinear transformation with a branch-cut at 2π as the activation functions in the input layer to map the real-valued features to the complex domain. The neurons in the hidden layer employ the fully complex-valued activation function of the type of a hyperbolic secant function. The parameters of the hidden layer are chosen randomly and the output weights are estimated as the minimum norm least square solution to a set of linear equations. The classification ability of these classifiers are evaluated using a set of benchmark data sets from the UCI machine learning repository. Results highlight the superior classification ability of these classifiers with least computational effort.

UR - http://www.scopus.com/inward/record.url?scp=79957799279&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79957799279&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-21105-8_70

DO - 10.1007/978-3-642-21105-8_70

M3 - Conference contribution

AN - SCOPUS:79957799279

SN - 9783642211041

VL - 6675 LNCS

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

SP - 602

EP - 609

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

ER -