A fully complex-valued radial basis function classifier for real-valued classification problems

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

Research output: Contribution to journalArticle

49 Citations (Scopus)

Abstract

In this paper, we investigate the decision making ability of a fully complex-valued radial basis function (FC-RBF) network in solving real-valued classification problems. The FC-RBF classifier is a single hidden layer fully complex-valued neural network with a nonlinear input layer, a nonlinear hidden layer, and a linear output layer. The neurons in the input layer of the classifier employ the phase encoded transformation to map the input features from the Real domain to the Complex domain. The neurons in the hidden layer employ a fully complex-valued Gaussian-like activation function of the type of hyperbolic secant (sech). The classification ability of the classifier is first studied analytically and it is shown that the decision boundaries of the FC-RBF classifier are orthogonal to each other. Then, the performance of the FC-RBF classifier is studied experimentally using a set of real-valued benchmark problems and also a real-world problem. The study clearly indicates the superior classification ability of the FC-RBF classifier.

Original languageEnglish
Pages (from-to)104-110
Number of pages7
JournalNeurocomputing
Volume78
Issue number1
DOIs
Publication statusPublished - 2012 Feb 15

Fingerprint

Aptitude
Classifiers
Neurons
Benchmarking
Decision Making
Radial basis function networks
Phase transitions
Decision making
Chemical activation
Neural networks

Keywords

  • Acoustic emission signal classification
  • Classification
  • Complex-valued classifiers
  • Fully complex-valued radial basis function
  • Orthogonal decision boundaries

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Cognitive Neuroscience

Cite this

A fully complex-valued radial basis function classifier for real-valued classification problems. / Savitha, R.; Suresh, S.; Sundararajan, N.; Kim, Hyong Joong.

In: Neurocomputing, Vol. 78, No. 1, 15.02.2012, p. 104-110.

Research output: Contribution to journalArticle

Savitha, R. ; Suresh, S. ; Sundararajan, N. ; Kim, Hyong Joong. / A fully complex-valued radial basis function classifier for real-valued classification problems. In: Neurocomputing. 2012 ; Vol. 78, No. 1. pp. 104-110.
@article{ef6801ce833a4e45a8bfe7ae31a55a2c,
title = "A fully complex-valued radial basis function classifier for real-valued classification problems",
abstract = "In this paper, we investigate the decision making ability of a fully complex-valued radial basis function (FC-RBF) network in solving real-valued classification problems. The FC-RBF classifier is a single hidden layer fully complex-valued neural network with a nonlinear input layer, a nonlinear hidden layer, and a linear output layer. The neurons in the input layer of the classifier employ the phase encoded transformation to map the input features from the Real domain to the Complex domain. The neurons in the hidden layer employ a fully complex-valued Gaussian-like activation function of the type of hyperbolic secant (sech). The classification ability of the classifier is first studied analytically and it is shown that the decision boundaries of the FC-RBF classifier are orthogonal to each other. Then, the performance of the FC-RBF classifier is studied experimentally using a set of real-valued benchmark problems and also a real-world problem. The study clearly indicates the superior classification ability of the FC-RBF classifier.",
keywords = "Acoustic emission signal classification, Classification, Complex-valued classifiers, Fully complex-valued radial basis function, Orthogonal decision boundaries",
author = "R. Savitha and S. Suresh and N. Sundararajan and Kim, {Hyong Joong}",
year = "2012",
month = "2",
day = "15",
doi = "10.1016/j.neucom.2011.05.036",
language = "English",
volume = "78",
pages = "104--110",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - A fully complex-valued radial basis function classifier for real-valued classification problems

AU - Savitha, R.

AU - Suresh, S.

AU - Sundararajan, N.

AU - Kim, Hyong Joong

PY - 2012/2/15

Y1 - 2012/2/15

N2 - In this paper, we investigate the decision making ability of a fully complex-valued radial basis function (FC-RBF) network in solving real-valued classification problems. The FC-RBF classifier is a single hidden layer fully complex-valued neural network with a nonlinear input layer, a nonlinear hidden layer, and a linear output layer. The neurons in the input layer of the classifier employ the phase encoded transformation to map the input features from the Real domain to the Complex domain. The neurons in the hidden layer employ a fully complex-valued Gaussian-like activation function of the type of hyperbolic secant (sech). The classification ability of the classifier is first studied analytically and it is shown that the decision boundaries of the FC-RBF classifier are orthogonal to each other. Then, the performance of the FC-RBF classifier is studied experimentally using a set of real-valued benchmark problems and also a real-world problem. The study clearly indicates the superior classification ability of the FC-RBF classifier.

AB - In this paper, we investigate the decision making ability of a fully complex-valued radial basis function (FC-RBF) network in solving real-valued classification problems. The FC-RBF classifier is a single hidden layer fully complex-valued neural network with a nonlinear input layer, a nonlinear hidden layer, and a linear output layer. The neurons in the input layer of the classifier employ the phase encoded transformation to map the input features from the Real domain to the Complex domain. The neurons in the hidden layer employ a fully complex-valued Gaussian-like activation function of the type of hyperbolic secant (sech). The classification ability of the classifier is first studied analytically and it is shown that the decision boundaries of the FC-RBF classifier are orthogonal to each other. Then, the performance of the FC-RBF classifier is studied experimentally using a set of real-valued benchmark problems and also a real-world problem. The study clearly indicates the superior classification ability of the FC-RBF classifier.

KW - Acoustic emission signal classification

KW - Classification

KW - Complex-valued classifiers

KW - Fully complex-valued radial basis function

KW - Orthogonal decision boundaries

UR - http://www.scopus.com/inward/record.url?scp=82655172397&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=82655172397&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2011.05.036

DO - 10.1016/j.neucom.2011.05.036

M3 - Article

AN - SCOPUS:82655172397

VL - 78

SP - 104

EP - 110

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

IS - 1

ER -