File Download
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1109/IJCNN.2013.6707063
- Scopus: eid_2-s2.0-84893554076
Supplementary
-
Citations:
- Scopus: 0
- Appears in Collections:
Conference Paper: Sparse similarity matrix learning for visual object retrieval
Title | Sparse similarity matrix learning for visual object retrieval |
---|---|
Authors | |
Keywords | Benchmark datasets Discriminability Object retrieval Off-diagonal elements Quantization errors Similarity matrix Similarity metrics TF-IDF weighting |
Issue Date | 2013 |
Publisher | IEEE. The Journal's web site is located at http://ieeexplore.ieee.org/xpl/conhome.jsp?punumber=1000500 |
Citation | The 2013 International Joint Conference on Neural Networks (IJCNN), Dallas, TX., 4-9 August 2013. In Conference Proceedings, 2013, p. 1-8 How to Cite? |
Abstract | Tf-idf weighting scheme is adopted by state-of-the-art object retrieval systems to reflect the difference in discriminability between visual words. However, we argue it is only suboptimal by noting that tf-idf weighting scheme does not take quantization error into account and exploit word correlation. We view tf-idf weights as an example of diagonal Mahalanobis-type similarity matrix and generalize it into a sparse one by selectively activating off-diagonal elements. Our goal is to separate similarity of relevant images from that of irrelevant ones by a safe margin. We satisfy such similarity constraints by learning an optimal similarity metric from labeled data. An effective scheme is developed to collect training data with an emphasis on cases where the tf-idf weights violates the relative relevance constraints. Experimental results on benchmark datasets indicate the learnt similarity metric consistently and significantly outperforms the tf-idf weighting scheme. © 2013 IEEE. |
Persistent Identifier | http://hdl.handle.net/10722/186495 |
ISBN |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Yan, Z | en_US |
dc.contributor.author | Yu, Y | en_US |
dc.date.accessioned | 2013-08-20T12:11:14Z | - |
dc.date.available | 2013-08-20T12:11:14Z | - |
dc.date.issued | 2013 | en_US |
dc.identifier.citation | The 2013 International Joint Conference on Neural Networks (IJCNN), Dallas, TX., 4-9 August 2013. In Conference Proceedings, 2013, p. 1-8 | en_US |
dc.identifier.isbn | 978-1-4673-6128-6 | - |
dc.identifier.uri | http://hdl.handle.net/10722/186495 | - |
dc.description.abstract | Tf-idf weighting scheme is adopted by state-of-the-art object retrieval systems to reflect the difference in discriminability between visual words. However, we argue it is only suboptimal by noting that tf-idf weighting scheme does not take quantization error into account and exploit word correlation. We view tf-idf weights as an example of diagonal Mahalanobis-type similarity matrix and generalize it into a sparse one by selectively activating off-diagonal elements. Our goal is to separate similarity of relevant images from that of irrelevant ones by a safe margin. We satisfy such similarity constraints by learning an optimal similarity metric from labeled data. An effective scheme is developed to collect training data with an emphasis on cases where the tf-idf weights violates the relative relevance constraints. Experimental results on benchmark datasets indicate the learnt similarity metric consistently and significantly outperforms the tf-idf weighting scheme. © 2013 IEEE. | - |
dc.language | eng | en_US |
dc.publisher | IEEE. The Journal's web site is located at http://ieeexplore.ieee.org/xpl/conhome.jsp?punumber=1000500 | - |
dc.relation.ispartof | International Joint Conference on Neural Networks (IJCNN) | en_US |
dc.subject | Benchmark datasets | - |
dc.subject | Discriminability | - |
dc.subject | Object retrieval | - |
dc.subject | Off-diagonal elements | - |
dc.subject | Quantization errors | - |
dc.subject | Similarity matrix | - |
dc.subject | Similarity metrics | - |
dc.subject | TF-IDF weighting | - |
dc.title | Sparse similarity matrix learning for visual object retrieval | en_US |
dc.type | Conference_Paper | en_US |
dc.identifier.email | Yu, Y: yzyu@cs.hku.hk | en_US |
dc.identifier.authority | Yu, Y=rp01415 | en_US |
dc.description.nature | link_to_subscribed_fulltext | - |
dc.identifier.doi | 10.1109/IJCNN.2013.6707063 | - |
dc.identifier.scopus | eid_2-s2.0-84893554076 | - |
dc.identifier.hkuros | 220947 | en_US |
dc.identifier.spage | 1 | - |
dc.identifier.epage | 8 | - |
dc.publisher.place | United States | - |
dc.customcontrol.immutable | sml 150204 | - |