A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition

Wensheng Chen; Pong Chi Yuen; Jian Huang; Daoqing Dai

doi:10.1007/978-3-540-30548-4_37

A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition

Wensheng Chen^*, Pong Chi Yuen, Jian Huang, Daoqing Dai

^*Corresponding author for this work

Department of Computer Science

Research output: Chapter in book/report/conference proceeding › Conference proceeding › peer-review

1 Citation (Scopus)

Abstract

In this paper, a new 1-parameter regularized discriminant analysis (IPRDA) algorithm is developed to deal with the small sample size (S3) problem. The main limitation in regularization is that the computational complexity of determining the optimal parameters is very high. In view of this limitation, we derive a single parameter (t) explicit expression formula for determining the 3 parameters. A simple and efficient method is proposed to determine the value of t. The proposed IPRLDA method for face recognition has been evaluated with two public available databases, namely ORL and FERET databases. The average recognition accuracy of 50 runs for ORL and FERET database are 96.65% and 94.00% respectively. Comparing with existing LDA-based methods in solving the S3 problem, the proposed 1PRLDA method gives the best performance.

Original language	English
Title of host publication	Advances in Biometric Person Authentication
Subtitle of host publication	5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004, Guangzhou, China, December 13-14, 2004, Proceedings
Editors	Stan Z. Li, Jianhuang Lai, Tieniu Tan, Guocan Feng, Yunhong Wang
Place of Publication	Berlin, Heidelberg
Publisher	Springer
Pages	320-329
Number of pages	10
Edition	1st
ISBN (Electronic)	9783540305484
ISBN (Print)	9783540240297
DOIs	https://doi.org/10.1007/978-3-540-30548-4_37
Publication status	Published - 29 Nov 2004
Event	5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004 - Guangzhou, China Duration: 13 Dec 2004 → 14 Dec 2004 https://link.springer.com/book/10.1007/b104239

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	3338
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Name	SINOBIOMETRICS: Chinese Conference on Biometric Recognition

Conference

Conference	5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004
Country/Territory	China
City	Guangzhou
Period	13/12/04 → 14/12/04
Internet address	https://link.springer.com/book/10.1007/b104239

User-Defined Keywords

Face Recognition
Linear Discriminant Analysis
Training Image
Recognition Accuracy
FERET Database

Access to Document

10.1007/978-3-540-30548-4_37

Cite this

Chen, W., Yuen, P. C., Huang, J., & Dai, D. (2004). A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition. In S. Z. Li, J. Lai, T. Tan, G. Feng, & Y. Wang (Eds.), Advances in Biometric Person Authentication: 5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004, Guangzhou, China, December 13-14, 2004, Proceedings (1st ed., pp. 320-329). (Lecture Notes in Computer Science; Vol. 3338), (SINOBIOMETRICS: Chinese Conference on Biometric Recognition). Springer. https://doi.org/10.1007/978-3-540-30548-4_37

Chen, Wensheng ; Yuen, Pong Chi ; Huang, Jian et al. / A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition. Advances in Biometric Person Authentication: 5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004, Guangzhou, China, December 13-14, 2004, Proceedings. editor / Stan Z. Li ; Jianhuang Lai ; Tieniu Tan ; Guocan Feng ; Yunhong Wang. 1st. ed. Berlin, Heidelberg : Springer, 2004. pp. 320-329 (Lecture Notes in Computer Science). (SINOBIOMETRICS: Chinese Conference on Biometric Recognition).

@inproceedings{c27adddbde2f4686bbff7bf7117baacb,

title = "A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition",

abstract = "In this paper, a new 1-parameter regularized discriminant analysis (IPRDA) algorithm is developed to deal with the small sample size (S3) problem. The main limitation in regularization is that the computational complexity of determining the optimal parameters is very high. In view of this limitation, we derive a single parameter (t) explicit expression formula for determining the 3 parameters. A simple and efficient method is proposed to determine the value of t. The proposed IPRLDA method for face recognition has been evaluated with two public available databases, namely ORL and FERET databases. The average recognition accuracy of 50 runs for ORL and FERET database are 96.65% and 94.00% respectively. Comparing with existing LDA-based methods in solving the S3 problem, the proposed 1PRLDA method gives the best performance.",

keywords = "Face Recognition, Linear Discriminant Analysis, Training Image, Recognition Accuracy, FERET Database",

author = "Wensheng Chen and Yuen, {Pong Chi} and Jian Huang and Daoqing Dai",

year = "2004",

month = nov,

day = "29",

doi = "10.1007/978-3-540-30548-4_37",

language = "English",

isbn = "9783540240297",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "320--329",

editor = "Li, {Stan Z.} and Jianhuang Lai and Tieniu Tan and Guocan Feng and Yunhong Wang",

booktitle = "Advances in Biometric Person Authentication",

edition = "1st",

url = "https://link.springer.com/book/10.1007/b104239",

}

Chen, W, Yuen, PC, Huang, J & Dai, D 2004, A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition. in SZ Li, J Lai, T Tan, G Feng & Y Wang (eds), Advances in Biometric Person Authentication: 5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004, Guangzhou, China, December 13-14, 2004, Proceedings. 1st edn, Lecture Notes in Computer Science, vol. 3338, SINOBIOMETRICS: Chinese Conference on Biometric Recognition, Springer, Berlin, Heidelberg, pp. 320-329, 5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004, Guangzhou, China, 13/12/04. https://doi.org/10.1007/978-3-540-30548-4_37

A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition. / Chen, Wensheng; Yuen, Pong Chi; Huang, Jian et al.
Advances in Biometric Person Authentication: 5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004, Guangzhou, China, December 13-14, 2004, Proceedings. ed. / Stan Z. Li; Jianhuang Lai; Tieniu Tan; Guocan Feng; Yunhong Wang. 1st. ed. Berlin, Heidelberg: Springer, 2004. p. 320-329 (Lecture Notes in Computer Science; Vol. 3338), (SINOBIOMETRICS: Chinese Conference on Biometric Recognition).

Research output: Chapter in book/report/conference proceeding › Conference proceeding › peer-review

TY - GEN

T1 - A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition

AU - Chen, Wensheng

AU - Yuen, Pong Chi

AU - Huang, Jian

AU - Dai, Daoqing

PY - 2004/11/29

Y1 - 2004/11/29

N2 - In this paper, a new 1-parameter regularized discriminant analysis (IPRDA) algorithm is developed to deal with the small sample size (S3) problem. The main limitation in regularization is that the computational complexity of determining the optimal parameters is very high. In view of this limitation, we derive a single parameter (t) explicit expression formula for determining the 3 parameters. A simple and efficient method is proposed to determine the value of t. The proposed IPRLDA method for face recognition has been evaluated with two public available databases, namely ORL and FERET databases. The average recognition accuracy of 50 runs for ORL and FERET database are 96.65% and 94.00% respectively. Comparing with existing LDA-based methods in solving the S3 problem, the proposed 1PRLDA method gives the best performance.

AB - In this paper, a new 1-parameter regularized discriminant analysis (IPRDA) algorithm is developed to deal with the small sample size (S3) problem. The main limitation in regularization is that the computational complexity of determining the optimal parameters is very high. In view of this limitation, we derive a single parameter (t) explicit expression formula for determining the 3 parameters. A simple and efficient method is proposed to determine the value of t. The proposed IPRLDA method for face recognition has been evaluated with two public available databases, namely ORL and FERET databases. The average recognition accuracy of 50 runs for ORL and FERET database are 96.65% and 94.00% respectively. Comparing with existing LDA-based methods in solving the S3 problem, the proposed 1PRLDA method gives the best performance.

KW - Face Recognition

KW - Linear Discriminant Analysis

KW - Training Image

KW - Recognition Accuracy

KW - FERET Database

UR - http://www.scopus.com/inward/record.url?scp=35048824807&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-30548-4_37

DO - 10.1007/978-3-540-30548-4_37

M3 - Conference proceeding

AN - SCOPUS:35048824807

SN - 9783540240297

T3 - Lecture Notes in Computer Science

SP - 320

EP - 329

BT - Advances in Biometric Person Authentication

A2 - Li, Stan Z.

A2 - Lai, Jianhuang

A2 - Tan, Tieniu

A2 - Feng, Guocan

A2 - Wang, Yunhong

PB - Springer

CY - Berlin, Heidelberg

T2 - 5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004

Y2 - 13 December 2004 through 14 December 2004

ER -

Chen W, Yuen PC, Huang J, Dai D. A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition. In Li SZ, Lai J, Tan T, Feng G, Wang Y, editors, Advances in Biometric Person Authentication: 5th Chinese Conference on Biometric Recognition, SINOBIOMETRICS 2004, Guangzhou, China, December 13-14, 2004, Proceedings. 1st ed. Berlin, Heidelberg: Springer. 2004. p. 320-329. (Lecture Notes in Computer Science). (SINOBIOMETRICS: Chinese Conference on Biometric Recognition). doi: 10.1007/978-3-540-30548-4_37

A Novel One-Parameter Regularized Linear Discriminant Analysis for Solving Small Sample Size Problem in Face Recognition

Abstract

Publication series

Conference

User-Defined Keywords

Access to Document

Other files and links

Fingerprint

Cite this