Gradient descent learning of nearest neighbor classifiers with outlier rejection

Kiichi Urahama, Y. Furukawa

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

The nearest neighbor classification rule is extended to reject outlier data and is implemented with an analog electronic circuit. A continuous membership function is derived from an optimization formulation of the classification rule. A learning algorithm is then presented for arranging prototype patterns to their optimal places and adjusting the radius of outlier rejection. The place of prototypes and the rejection radius are incrementally updated at every presentation of training patterns in the steepest descent direction of the error of the membership of the presented pattern from its correct value. Some elementary experiments examplify the convergence of the present learning algorithm.

Original languageEnglish
Pages (from-to)761-768
Number of pages8
JournalPattern Recognition
Volume28
Issue number5
DOIs
Publication statusPublished - Jan 1 1995

Fingerprint

Learning algorithms
Classifiers
Membership functions
Networks (circuits)
Experiments

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Computer Vision and Pattern Recognition
  • Signal Processing
  • Software
  • Electrical and Electronic Engineering

Cite this

Gradient descent learning of nearest neighbor classifiers with outlier rejection. / Urahama, Kiichi; Furukawa, Y.

In: Pattern Recognition, Vol. 28, No. 5, 01.01.1995, p. 761-768.

Research output: Contribution to journalArticle

Urahama, Kiichi ; Furukawa, Y. / Gradient descent learning of nearest neighbor classifiers with outlier rejection. In: Pattern Recognition. 1995 ; Vol. 28, No. 5. pp. 761-768.
@article{439e04cdf7734a43aed7bcb48aa00b2b,
title = "Gradient descent learning of nearest neighbor classifiers with outlier rejection",
abstract = "The nearest neighbor classification rule is extended to reject outlier data and is implemented with an analog electronic circuit. A continuous membership function is derived from an optimization formulation of the classification rule. A learning algorithm is then presented for arranging prototype patterns to their optimal places and adjusting the radius of outlier rejection. The place of prototypes and the rejection radius are incrementally updated at every presentation of training patterns in the steepest descent direction of the error of the membership of the presented pattern from its correct value. Some elementary experiments examplify the convergence of the present learning algorithm.",
author = "Kiichi Urahama and Y. Furukawa",
year = "1995",
month = "1",
day = "1",
doi = "10.1016/0031-3203(94)00142-9",
language = "English",
volume = "28",
pages = "761--768",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",
number = "5",

}

TY - JOUR

T1 - Gradient descent learning of nearest neighbor classifiers with outlier rejection

AU - Urahama, Kiichi

AU - Furukawa, Y.

PY - 1995/1/1

Y1 - 1995/1/1

N2 - The nearest neighbor classification rule is extended to reject outlier data and is implemented with an analog electronic circuit. A continuous membership function is derived from an optimization formulation of the classification rule. A learning algorithm is then presented for arranging prototype patterns to their optimal places and adjusting the radius of outlier rejection. The place of prototypes and the rejection radius are incrementally updated at every presentation of training patterns in the steepest descent direction of the error of the membership of the presented pattern from its correct value. Some elementary experiments examplify the convergence of the present learning algorithm.

AB - The nearest neighbor classification rule is extended to reject outlier data and is implemented with an analog electronic circuit. A continuous membership function is derived from an optimization formulation of the classification rule. A learning algorithm is then presented for arranging prototype patterns to their optimal places and adjusting the radius of outlier rejection. The place of prototypes and the rejection radius are incrementally updated at every presentation of training patterns in the steepest descent direction of the error of the membership of the presented pattern from its correct value. Some elementary experiments examplify the convergence of the present learning algorithm.

UR - http://www.scopus.com/inward/record.url?scp=0029307293&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0029307293&partnerID=8YFLogxK

U2 - 10.1016/0031-3203(94)00142-9

DO - 10.1016/0031-3203(94)00142-9

M3 - Article

VL - 28

SP - 761

EP - 768

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

IS - 5

ER -