Semi-Paired Asymmetric Deep Cross-Modal Hashing Learning
Hashing has been effectively applied in large-scale multimedia retrieval tasks due to the characteristics of fast calculation speed and low storage cost. However, most existing cross-view hashing methods require well paired views, where one sample in one view can always be associated with one sample...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9119427/ |
| _version_ | 1818876634784071680 |
|---|---|
| author | Yi Wang Xiaobo Shen Zhenmin Tang Tao Zhang Jianyong Lv |
| author_facet | Yi Wang Xiaobo Shen Zhenmin Tang Tao Zhang Jianyong Lv |
| author_sort | Yi Wang |
| collection | DOAJ |
| description | Hashing has been effectively applied in large-scale multimedia retrieval tasks due to the characteristics of fast calculation speed and low storage cost. However, most existing cross-view hashing methods require well paired views, where one sample in one view can always be associated with one sample in another view. Nevertheless, such fully-paired setting is difficult to hold in practice. The cross-view retrieval task for semi-paired data, in which only a portion of the sample is paired, remains more challenging and less explored. In this case, we design an unsupervised deep hashing for large-scale data, named semi-paired asymmetric deep cross-model hashing (SADCH) to address the challenging task. SADCH is a novel asymmetric end-to-end deep neural network model. Specifically, SADCH trains deep network by using query points to improve the training efficiency, and directly learns hash codes of database. A similarity matrix is constructed by a novel cross-view graph to explore the underlying data structures, such that the similarities of paired points and unpaired points can be preserved. As such, the deep features and similarity graph matrix can be jointly used to design an alternating algorithm to efficiently generate more differentiated hash codes. SADCH is evaluated for large-scale cross-view approximate nearest neighbor search on three benchmark datasets and compared against several state-of-the arts. The results of extensive experiments demonstrate the superiority of the proposed SADCH for semi-paired hashing in the unsupervised setting. |
| first_indexed | 2024-12-19T13:45:31Z |
| format | Article |
| id | doaj.art-6973cea2a0d340968bf1fdcc8738e1b7 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-19T13:45:31Z |
| publishDate | 2020-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-6973cea2a0d340968bf1fdcc8738e1b72022-12-21T20:18:53ZengIEEEIEEE Access2169-35362020-01-01811381411382510.1109/ACCESS.2020.30030729119427Semi-Paired Asymmetric Deep Cross-Modal Hashing LearningYi Wang0https://orcid.org/0000-0002-2578-7693Xiaobo Shen1https://orcid.org/0000-0001-8494-4532Zhenmin Tang2https://orcid.org/0000-0002-2889-796XTao Zhang3https://orcid.org/0000-0002-8359-3387Jianyong Lv4https://orcid.org/0000-0003-1438-0078School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, ChinaSchool of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, ChinaSchool of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, ChinaCollege of Information Engineering, Yangzhou University, Yangzhou, ChinaSchool of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, ChinaHashing has been effectively applied in large-scale multimedia retrieval tasks due to the characteristics of fast calculation speed and low storage cost. However, most existing cross-view hashing methods require well paired views, where one sample in one view can always be associated with one sample in another view. Nevertheless, such fully-paired setting is difficult to hold in practice. The cross-view retrieval task for semi-paired data, in which only a portion of the sample is paired, remains more challenging and less explored. In this case, we design an unsupervised deep hashing for large-scale data, named semi-paired asymmetric deep cross-model hashing (SADCH) to address the challenging task. SADCH is a novel asymmetric end-to-end deep neural network model. Specifically, SADCH trains deep network by using query points to improve the training efficiency, and directly learns hash codes of database. A similarity matrix is constructed by a novel cross-view graph to explore the underlying data structures, such that the similarities of paired points and unpaired points can be preserved. As such, the deep features and similarity graph matrix can be jointly used to design an alternating algorithm to efficiently generate more differentiated hash codes. SADCH is evaluated for large-scale cross-view approximate nearest neighbor search on three benchmark datasets and compared against several state-of-the arts. The results of extensive experiments demonstrate the superiority of the proposed SADCH for semi-paired hashing in the unsupervised setting.https://ieeexplore.ieee.org/document/9119427/Deep hashingcross-view retrievalasymmetric learningsemi-paired data |
| spellingShingle | Yi Wang Xiaobo Shen Zhenmin Tang Tao Zhang Jianyong Lv Semi-Paired Asymmetric Deep Cross-Modal Hashing Learning IEEE Access Deep hashing cross-view retrieval asymmetric learning semi-paired data |
| title | Semi-Paired Asymmetric Deep Cross-Modal Hashing Learning |
| title_full | Semi-Paired Asymmetric Deep Cross-Modal Hashing Learning |
| title_fullStr | Semi-Paired Asymmetric Deep Cross-Modal Hashing Learning |
| title_full_unstemmed | Semi-Paired Asymmetric Deep Cross-Modal Hashing Learning |
| title_short | Semi-Paired Asymmetric Deep Cross-Modal Hashing Learning |
| title_sort | semi paired asymmetric deep cross modal hashing learning |
| topic | Deep hashing cross-view retrieval asymmetric learning semi-paired data |
| url | https://ieeexplore.ieee.org/document/9119427/ |
| work_keys_str_mv | AT yiwang semipairedasymmetricdeepcrossmodalhashinglearning AT xiaoboshen semipairedasymmetricdeepcrossmodalhashinglearning AT zhenmintang semipairedasymmetricdeepcrossmodalhashinglearning AT taozhang semipairedasymmetricdeepcrossmodalhashinglearning AT jianyonglv semipairedasymmetricdeepcrossmodalhashinglearning |