Abstract
A new formulation of metric learning is introduced by assimilating the kernel ridge regression (KRR) and weighted side-information linear discriminant analysis (WSILD) to enjoy the best of both worlds for unconstrained face verification task. To be specific, we formulate a doublet constrained metric learning problem by means of a second degree polynomial kernel function. The said metric learning problem can be solved analytically for Mahalanobis distance metric due to simplistic nature of KRR in which we named KRRML. In addition, the WSILD further enhances the learned Mahalanobis distance metric by leveraging the within-class and between-class scatter matrix of doublets. We evaluate the proposed method with Labeled Faces in the Wild database, a large benchmark dataset targeted for unconstrained face verification. The promising result attests the robustness and feasibility of the proposed method.
| Original language | English |
|---|---|
| Title of host publication | Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings |
| Editors | Seiichi Ozawa, Kazushi Ikeda, Derong Liu, Akira Hirose, Kenji Doya, Minho Lee |
| Publisher | Springer Verlag |
| Pages | 401-410 |
| Number of pages | 10 |
| ISBN (Print) | 9783319466712 |
| DOIs | |
| Publication status | Published - 2016 Jan 1 |
| Event | 23rd International Conference on Neural Information Processing, ICONIP 2016 - Kyoto, Japan Duration: 2016 Oct 16 → 2016 Oct 21 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Volume | 9948 LNCS |
| ISSN (Print) | 0302-9743 |
| ISSN (Electronic) | 1611-3349 |
Other
| Other | 23rd International Conference on Neural Information Processing, ICONIP 2016 |
|---|---|
| Country | Japan |
| City | Kyoto |
| Period | 16/10/16 → 16/10/21 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Computer Science(all)
Cite this
}
Weighted discriminant analysis and kernel ridge regression metric learning for face verification. / Chong, Siew Chin; Teoh, Andrew Beng Jin; Ong, Thian Song.
Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings. ed. / Seiichi Ozawa; Kazushi Ikeda; Derong Liu; Akira Hirose; Kenji Doya; Minho Lee. Springer Verlag, 2016. p. 401-410 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9948 LNCS).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
TY - GEN
T1 - Weighted discriminant analysis and kernel ridge regression metric learning for face verification
AU - Chong, Siew Chin
AU - Teoh, Andrew Beng Jin
AU - Ong, Thian Song
PY - 2016/1/1
Y1 - 2016/1/1
N2 - A new formulation of metric learning is introduced by assimilating the kernel ridge regression (KRR) and weighted side-information linear discriminant analysis (WSILD) to enjoy the best of both worlds for unconstrained face verification task. To be specific, we formulate a doublet constrained metric learning problem by means of a second degree polynomial kernel function. The said metric learning problem can be solved analytically for Mahalanobis distance metric due to simplistic nature of KRR in which we named KRRML. In addition, the WSILD further enhances the learned Mahalanobis distance metric by leveraging the within-class and between-class scatter matrix of doublets. We evaluate the proposed method with Labeled Faces in the Wild database, a large benchmark dataset targeted for unconstrained face verification. The promising result attests the robustness and feasibility of the proposed method.
AB - A new formulation of metric learning is introduced by assimilating the kernel ridge regression (KRR) and weighted side-information linear discriminant analysis (WSILD) to enjoy the best of both worlds for unconstrained face verification task. To be specific, we formulate a doublet constrained metric learning problem by means of a second degree polynomial kernel function. The said metric learning problem can be solved analytically for Mahalanobis distance metric due to simplistic nature of KRR in which we named KRRML. In addition, the WSILD further enhances the learned Mahalanobis distance metric by leveraging the within-class and between-class scatter matrix of doublets. We evaluate the proposed method with Labeled Faces in the Wild database, a large benchmark dataset targeted for unconstrained face verification. The promising result attests the robustness and feasibility of the proposed method.
UR - http://www.scopus.com/inward/record.url?scp=84992553675&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84992553675&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-46672-9_45
DO - 10.1007/978-3-319-46672-9_45
M3 - Conference contribution
AN - SCOPUS:84992553675
SN - 9783319466712
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 401
EP - 410
BT - Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings
A2 - Ozawa, Seiichi
A2 - Ikeda, Kazushi
A2 - Liu, Derong
A2 - Hirose, Akira
A2 - Doya, Kenji
A2 - Lee, Minho
PB - Springer Verlag
ER -