A Novel Metric Learning Method and Its Application

Author

Khadir, A. Shaik Abdul ; Raja, S. V Kasmir ; Amanullah, K. Mohamed

Author_Institution

Dept. of Comp. Sci., Khadir Mohideen Coll., Adirampattinam, India

fYear

2009

fDate

28-29 Dec. 2009

Firstpage

827

Lastpage

829

Abstract

In this paper, we propose a novel metric learning method based on regularized moving least squares. Unlike most previous metric learning methods which learn a global Mahalanobis distance, we define locally smooth metrics using local affine transformations which are more flexible. Experimental results provide empirical evidence for the effectiveness of our approach.

Keywords

content-based retrieval; learning (artificial intelligence); least squares approximations; content-based image retrieval; global Mahalanobis distance; local affine transformations; locally smooth metrics; metric learning method; nonlinear dimensionality reduction; regularized moving least squares; topological structures; Clustering algorithms; Content based retrieval; Educational institutions; Image retrieval; Learning systems; Least squares methods; Machine learning; Optimization methods; Semisupervised learning; Telecommunication computing; Content based image retrieval; Mehalanobis metrics; Semi supervised learning;

fLanguage

English

Publisher

ieee

Conference_Titel

Advances in Computing, Control, & Telecommunication Technologies, 2009. ACT '09. International Conference on

Conference_Location

Trivandrum, Kerala

Print_ISBN

978-1-4244-5321-4

Electronic_ISBN

978-0-7695-3915-7

Type

conf

DOI

10.1109/ACT.2009.209

Filename

5375773