Locality sensitive discriminant analysis

Deng Cai, Xiaofei He, Kun Zhou, Jiawei Han, Hujun Bao

Research output: Contribution to journalConference article

Abstract

Linear Discriminant Analysis (LDA) is a popular data-analytic tool for studying the class relationship between data points. A major disadvantage of LDA is that it fails to discover the local geometrical structure of the data manifold. In this paper, we introduce a novel linear algorithm for discriminant analysis, called Locality Sensitive Discriminant Analysis (LSDA).When there is no sufficient training samples, local structure is generally more important than global structure for discriminant analysis. By discovering the local manifold structure, LSDA finds a projection which maximizes the margin between data points from different classes at each local area. Specifically, the data points are mapped into a subspace in which the nearby points with the same label are close to each other while the nearby points with different labels are far apart. Experiments carried out on several standard face databases show a clear improvement over the results of LDA-based recognition.

Original languageEnglish (US)
Pages (from-to)708-713
Number of pages6
JournalIJCAI International Joint Conference on Artificial Intelligence
StatePublished - Dec 1 2007
Event20th International Joint Conference on Artificial Intelligence, IJCAI 2007 - Hyderabad, India
Duration: Jan 6 2007Jan 12 2007

Fingerprint

Discriminant analysis
Labels
Experiments

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Locality sensitive discriminant analysis. / Cai, Deng; He, Xiaofei; Zhou, Kun; Han, Jiawei; Bao, Hujun.

In: IJCAI International Joint Conference on Artificial Intelligence, 01.12.2007, p. 708-713.

Research output: Contribution to journalConference article

Cai, Deng ; He, Xiaofei ; Zhou, Kun ; Han, Jiawei ; Bao, Hujun. / Locality sensitive discriminant analysis. In: IJCAI International Joint Conference on Artificial Intelligence. 2007 ; pp. 708-713.
@article{1f1319d1632c4059a350f90293990b38,
title = "Locality sensitive discriminant analysis",
abstract = "Linear Discriminant Analysis (LDA) is a popular data-analytic tool for studying the class relationship between data points. A major disadvantage of LDA is that it fails to discover the local geometrical structure of the data manifold. In this paper, we introduce a novel linear algorithm for discriminant analysis, called Locality Sensitive Discriminant Analysis (LSDA).When there is no sufficient training samples, local structure is generally more important than global structure for discriminant analysis. By discovering the local manifold structure, LSDA finds a projection which maximizes the margin between data points from different classes at each local area. Specifically, the data points are mapped into a subspace in which the nearby points with the same label are close to each other while the nearby points with different labels are far apart. Experiments carried out on several standard face databases show a clear improvement over the results of LDA-based recognition.",
author = "Deng Cai and Xiaofei He and Kun Zhou and Jiawei Han and Hujun Bao",
year = "2007",
month = "12",
day = "1",
language = "English (US)",
pages = "708--713",
journal = "IJCAI International Joint Conference on Artificial Intelligence",
issn = "1045-0823",

}

TY - JOUR

T1 - Locality sensitive discriminant analysis

AU - Cai, Deng

AU - He, Xiaofei

AU - Zhou, Kun

AU - Han, Jiawei

AU - Bao, Hujun

PY - 2007/12/1

Y1 - 2007/12/1

N2 - Linear Discriminant Analysis (LDA) is a popular data-analytic tool for studying the class relationship between data points. A major disadvantage of LDA is that it fails to discover the local geometrical structure of the data manifold. In this paper, we introduce a novel linear algorithm for discriminant analysis, called Locality Sensitive Discriminant Analysis (LSDA).When there is no sufficient training samples, local structure is generally more important than global structure for discriminant analysis. By discovering the local manifold structure, LSDA finds a projection which maximizes the margin between data points from different classes at each local area. Specifically, the data points are mapped into a subspace in which the nearby points with the same label are close to each other while the nearby points with different labels are far apart. Experiments carried out on several standard face databases show a clear improvement over the results of LDA-based recognition.

AB - Linear Discriminant Analysis (LDA) is a popular data-analytic tool for studying the class relationship between data points. A major disadvantage of LDA is that it fails to discover the local geometrical structure of the data manifold. In this paper, we introduce a novel linear algorithm for discriminant analysis, called Locality Sensitive Discriminant Analysis (LSDA).When there is no sufficient training samples, local structure is generally more important than global structure for discriminant analysis. By discovering the local manifold structure, LSDA finds a projection which maximizes the margin between data points from different classes at each local area. Specifically, the data points are mapped into a subspace in which the nearby points with the same label are close to each other while the nearby points with different labels are far apart. Experiments carried out on several standard face databases show a clear improvement over the results of LDA-based recognition.

UR - http://www.scopus.com/inward/record.url?scp=84880899766&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84880899766&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:84880899766

SP - 708

EP - 713

JO - IJCAI International Joint Conference on Artificial Intelligence

JF - IJCAI International Joint Conference on Artificial Intelligence

SN - 1045-0823

ER -