Nonlinear discriminant adaptive nearest neighbor classifiers

Peng Zhang, Jing Peng, S. Richard F Sims

Research output: Contribution to journalConference articleResearchpeer-review

Abstract

Nearest neighbor classifiers are one of most common techniques for classification and ATR applications. Hastie and Tibshirani propose a discriminant adaptive nearest neighbor (DANN) rule for computing a distance metric locally so that posterior probabilities tend to be homogeneous in the modified neighborhoods. The idea is to enlongate or constrict the neighborhood along the direction that is parallel or perpendicular to the decision boundary between two classes. DANN morphs a neighborhood in a linear fashion. In this paper, we extend it to the nonlinear case using the kernel trick. We demonstrate the efficacy of our kernel DANN in the context of ATR applications using a number of data sets.

Original languageEnglish
Article number40
Pages (from-to)359-369
Number of pages11
JournalProceedings of SPIE - The International Society for Optical Engineering
Volume5807
DOIs
StatePublished - 10 Nov 2005
EventAutomatic Target Recognition XV - Orlando, FL, United States
Duration: 29 Mar 200531 Mar 2005

Fingerprint

classifiers
Discriminant
Nearest Neighbor
Classifiers
Classifier
kernel
Distance Metric
Posterior Probability
Perpendicular
Efficacy
Tend
Computing
Demonstrate
Direction compound

Keywords

  • ATR
  • Kernel methods
  • Nearest neighbors

Cite this

@article{9563483c8a07410dbebcf1f272c98178,
title = "Nonlinear discriminant adaptive nearest neighbor classifiers",
abstract = "Nearest neighbor classifiers are one of most common techniques for classification and ATR applications. Hastie and Tibshirani propose a discriminant adaptive nearest neighbor (DANN) rule for computing a distance metric locally so that posterior probabilities tend to be homogeneous in the modified neighborhoods. The idea is to enlongate or constrict the neighborhood along the direction that is parallel or perpendicular to the decision boundary between two classes. DANN morphs a neighborhood in a linear fashion. In this paper, we extend it to the nonlinear case using the kernel trick. We demonstrate the efficacy of our kernel DANN in the context of ATR applications using a number of data sets.",
keywords = "ATR, Kernel methods, Nearest neighbors",
author = "Peng Zhang and Jing Peng and {Richard F Sims}, S.",
year = "2005",
month = "11",
day = "10",
doi = "10.1117/12.604150",
language = "English",
volume = "5807",
pages = "359--369",
journal = "Proceedings of SPIE - The International Society for Optical Engineering",
issn = "0277-786X",
publisher = "SPIE",

}

Nonlinear discriminant adaptive nearest neighbor classifiers. / Zhang, Peng; Peng, Jing; Richard F Sims, S.

In: Proceedings of SPIE - The International Society for Optical Engineering, Vol. 5807, 40, 10.11.2005, p. 359-369.

Research output: Contribution to journalConference articleResearchpeer-review

TY - JOUR

T1 - Nonlinear discriminant adaptive nearest neighbor classifiers

AU - Zhang, Peng

AU - Peng, Jing

AU - Richard F Sims, S.

PY - 2005/11/10

Y1 - 2005/11/10

N2 - Nearest neighbor classifiers are one of most common techniques for classification and ATR applications. Hastie and Tibshirani propose a discriminant adaptive nearest neighbor (DANN) rule for computing a distance metric locally so that posterior probabilities tend to be homogeneous in the modified neighborhoods. The idea is to enlongate or constrict the neighborhood along the direction that is parallel or perpendicular to the decision boundary between two classes. DANN morphs a neighborhood in a linear fashion. In this paper, we extend it to the nonlinear case using the kernel trick. We demonstrate the efficacy of our kernel DANN in the context of ATR applications using a number of data sets.

AB - Nearest neighbor classifiers are one of most common techniques for classification and ATR applications. Hastie and Tibshirani propose a discriminant adaptive nearest neighbor (DANN) rule for computing a distance metric locally so that posterior probabilities tend to be homogeneous in the modified neighborhoods. The idea is to enlongate or constrict the neighborhood along the direction that is parallel or perpendicular to the decision boundary between two classes. DANN morphs a neighborhood in a linear fashion. In this paper, we extend it to the nonlinear case using the kernel trick. We demonstrate the efficacy of our kernel DANN in the context of ATR applications using a number of data sets.

KW - ATR

KW - Kernel methods

KW - Nearest neighbors

UR - http://www.scopus.com/inward/record.url?scp=27544452969&partnerID=8YFLogxK

U2 - 10.1117/12.604150

DO - 10.1117/12.604150

M3 - Conference article

VL - 5807

SP - 359

EP - 369

JO - Proceedings of SPIE - The International Society for Optical Engineering

JF - Proceedings of SPIE - The International Society for Optical Engineering

SN - 0277-786X

M1 - 40

ER -