Analytical decision boundary feature extraction for neural networks

Jinwook Go, Chulhee Lee

Research output: Contribution to conferencePaper

5 Citations (Scopus)

Abstract

Recently, a feature extraction method based on decision boundary has been proposed for neural networks. The method is based on the fact that the vector normal to the decision boundary contains information useful for discriminating between classes. However, the normal vector was estimated numerically, resulting in inaccurate estimation and a long computational time. In this paper, we propose a new method to calculate the normal vector analytically and derive all the necessary equations for 3 layer feedforward neural networks with a sigmoid function. Experiments show that the proposed method provides a noticeably improved performance.

Original languageEnglish
Pages3072-3074
Number of pages3
Publication statusPublished - 2000 Dec 1
Event2000 International Geoscience and Remote Sensing Symposium (IGARSS 2000) - Honolulu, HI, USA
Duration: 2000 Jul 242000 Jul 28

Other

Other2000 International Geoscience and Remote Sensing Symposium (IGARSS 2000)
CityHonolulu, HI, USA
Period00/7/2400/7/28

Fingerprint

Feature extraction
Neural networks
Feedforward neural networks
extraction method
decision
method
experiment
Experiments

All Science Journal Classification (ASJC) codes

  • Software
  • Geology

Cite this

Go, J., & Lee, C. (2000). Analytical decision boundary feature extraction for neural networks. 3072-3074. Paper presented at 2000 International Geoscience and Remote Sensing Symposium (IGARSS 2000), Honolulu, HI, USA, .
Go, Jinwook ; Lee, Chulhee. / Analytical decision boundary feature extraction for neural networks. Paper presented at 2000 International Geoscience and Remote Sensing Symposium (IGARSS 2000), Honolulu, HI, USA, .3 p.
@conference{c96d75c8753742388a231850bacd43cd,
title = "Analytical decision boundary feature extraction for neural networks",
abstract = "Recently, a feature extraction method based on decision boundary has been proposed for neural networks. The method is based on the fact that the vector normal to the decision boundary contains information useful for discriminating between classes. However, the normal vector was estimated numerically, resulting in inaccurate estimation and a long computational time. In this paper, we propose a new method to calculate the normal vector analytically and derive all the necessary equations for 3 layer feedforward neural networks with a sigmoid function. Experiments show that the proposed method provides a noticeably improved performance.",
author = "Jinwook Go and Chulhee Lee",
year = "2000",
month = "12",
day = "1",
language = "English",
pages = "3072--3074",
note = "2000 International Geoscience and Remote Sensing Symposium (IGARSS 2000) ; Conference date: 24-07-2000 Through 28-07-2000",

}

Go, J & Lee, C 2000, 'Analytical decision boundary feature extraction for neural networks', Paper presented at 2000 International Geoscience and Remote Sensing Symposium (IGARSS 2000), Honolulu, HI, USA, 00/7/24 - 00/7/28 pp. 3072-3074.

Analytical decision boundary feature extraction for neural networks. / Go, Jinwook; Lee, Chulhee.

2000. 3072-3074 Paper presented at 2000 International Geoscience and Remote Sensing Symposium (IGARSS 2000), Honolulu, HI, USA, .

Research output: Contribution to conferencePaper

TY - CONF

T1 - Analytical decision boundary feature extraction for neural networks

AU - Go, Jinwook

AU - Lee, Chulhee

PY - 2000/12/1

Y1 - 2000/12/1

N2 - Recently, a feature extraction method based on decision boundary has been proposed for neural networks. The method is based on the fact that the vector normal to the decision boundary contains information useful for discriminating between classes. However, the normal vector was estimated numerically, resulting in inaccurate estimation and a long computational time. In this paper, we propose a new method to calculate the normal vector analytically and derive all the necessary equations for 3 layer feedforward neural networks with a sigmoid function. Experiments show that the proposed method provides a noticeably improved performance.

AB - Recently, a feature extraction method based on decision boundary has been proposed for neural networks. The method is based on the fact that the vector normal to the decision boundary contains information useful for discriminating between classes. However, the normal vector was estimated numerically, resulting in inaccurate estimation and a long computational time. In this paper, we propose a new method to calculate the normal vector analytically and derive all the necessary equations for 3 layer feedforward neural networks with a sigmoid function. Experiments show that the proposed method provides a noticeably improved performance.

UR - http://www.scopus.com/inward/record.url?scp=0034546602&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0034546602&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:0034546602

SP - 3072

EP - 3074

ER -

Go J, Lee C. Analytical decision boundary feature extraction for neural networks. 2000. Paper presented at 2000 International Geoscience and Remote Sensing Symposium (IGARSS 2000), Honolulu, HI, USA, .