Constrained empirical risk minimization framework for distance metric learning
Distance metric learning (DML) has received increasing attention in recent years. In this paper, we propose a constrained empirical risk minimization framework for DML. This framework enriches the state-of-the-art studies on both theoretic and algorithmic aspects. Theoretically, we comprehensively a...
Main Authors: | , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | unknown |
Published: |
2012
|
Subjects: | |
Online Access: | http://hdl.handle.net/10453/22849 |
id |
ftunivtsydney:oai:opus.lib.uts.edu.au:10453/22849 |
---|---|
record_format |
openpolar |
spelling |
ftunivtsydney:oai:opus.lib.uts.edu.au:10453/22849 2023-05-15T16:01:20+02:00 Constrained empirical risk minimization framework for distance metric learning Bian, W Tao, D 2012-12-01 application/pdf http://hdl.handle.net/10453/22849 unknown IEEE Transactions on Neural Networks and Learning Systems 10.1109/TNNLS.2012.2198075 IEEE Transactions on Neural Networks and Learning Systems, 2012, 23 (8), pp. 1194 - 1205 2162-237X http://hdl.handle.net/10453/22849 Artificial Intelligence & Image Processing Journal Article 2012 ftunivtsydney 2022-03-13T13:57:53Z Distance metric learning (DML) has received increasing attention in recent years. In this paper, we propose a constrained empirical risk minimization framework for DML. This framework enriches the state-of-the-art studies on both theoretic and algorithmic aspects. Theoretically, we comprehensively analyze the generalization by bounding the sample and the approximation errors with respect to the best model. Algorithmically, we carefully derive an optimal gradient descent by using Nesterov's method, and provide two example algorithms that utilize the logarithmic loss and the smoothed hinge loss, respectively. We evaluate the new framework on data classification and image retrieval experiments. Results show that the new framework has competitive performance compared with the representative DML algorithms, including Xing's method, large margin nearest neighbor classifier, neighborhood component analysis, and regularized metric learning. © 2012 IEEE. Article in Journal/Newspaper DML University of Technology Sydney: OPUS - Open Publications of UTS Scholars |
institution |
Open Polar |
collection |
University of Technology Sydney: OPUS - Open Publications of UTS Scholars |
op_collection_id |
ftunivtsydney |
language |
unknown |
topic |
Artificial Intelligence & Image Processing |
spellingShingle |
Artificial Intelligence & Image Processing Bian, W Tao, D Constrained empirical risk minimization framework for distance metric learning |
topic_facet |
Artificial Intelligence & Image Processing |
description |
Distance metric learning (DML) has received increasing attention in recent years. In this paper, we propose a constrained empirical risk minimization framework for DML. This framework enriches the state-of-the-art studies on both theoretic and algorithmic aspects. Theoretically, we comprehensively analyze the generalization by bounding the sample and the approximation errors with respect to the best model. Algorithmically, we carefully derive an optimal gradient descent by using Nesterov's method, and provide two example algorithms that utilize the logarithmic loss and the smoothed hinge loss, respectively. We evaluate the new framework on data classification and image retrieval experiments. Results show that the new framework has competitive performance compared with the representative DML algorithms, including Xing's method, large margin nearest neighbor classifier, neighborhood component analysis, and regularized metric learning. © 2012 IEEE. |
format |
Article in Journal/Newspaper |
author |
Bian, W Tao, D |
author_facet |
Bian, W Tao, D |
author_sort |
Bian, W |
title |
Constrained empirical risk minimization framework for distance metric learning |
title_short |
Constrained empirical risk minimization framework for distance metric learning |
title_full |
Constrained empirical risk minimization framework for distance metric learning |
title_fullStr |
Constrained empirical risk minimization framework for distance metric learning |
title_full_unstemmed |
Constrained empirical risk minimization framework for distance metric learning |
title_sort |
constrained empirical risk minimization framework for distance metric learning |
publishDate |
2012 |
url |
http://hdl.handle.net/10453/22849 |
genre |
DML |
genre_facet |
DML |
op_relation |
IEEE Transactions on Neural Networks and Learning Systems 10.1109/TNNLS.2012.2198075 IEEE Transactions on Neural Networks and Learning Systems, 2012, 23 (8), pp. 1194 - 1205 2162-237X http://hdl.handle.net/10453/22849 |
_version_ |
1766397243743535104 |