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.
|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|
|Number of pages||10|
|Publication status||Published - 2016|
|Event||23rd International Conference on Neural Information Processing, ICONIP 2016 - Kyoto, Japan|
Duration: 2016 Oct 16 → 2016 Oct 21
|Name||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Other||23rd International Conference on Neural Information Processing, ICONIP 2016|
|Period||16/10/16 → 16/10/21|
Bibliographical noteFunding Information:
The authors would like to thank Malaysia’s Fundamental Research Grant Scheme for supporting the research under grants MMUE/140026.
© Springer International Publishing AG 2016.
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Computer Science(all)