Kernelised reference‐wise metric learning
Unlike the doublet or triplet constraints, a novel kernelised reference‐wise metric learning is proposed by constructing reference‐wise constraints, which contain similarity information of each sample to all reference samples. After selecting several representative training samples as the reference...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2017-03-01
|
| Series: | Electronics Letters |
| Subjects: | |
| Online Access: | https://doi.org/10.1049/el.2016.4079 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832096664412749824 |
|---|---|
| author | Meng Wu Kai Luo Daijin Li Jun Zhou |
| author_facet | Meng Wu Kai Luo Daijin Li Jun Zhou |
| author_sort | Meng Wu |
| collection | DOAJ |
| description | Unlike the doublet or triplet constraints, a novel kernelised reference‐wise metric learning is proposed by constructing reference‐wise constraints, which contain similarity information of each sample to all reference samples. After selecting several representative training samples as the reference set, the training data are first mapped into a projected space by the reference sample matrix. Then the problem of metric learning is casted as a multi‐label classification problem under the reference‐wise constraints, in which an l2‐norm regularised least squares canonical correlation analysis can be used. Besides, the formulation is generalised to kernelised version for further boosting the performance. Experiments on two benchmark person re‐identification datasets demonstrate that the approach clearly outperforms the state‐of‐the‐art methods. |
| format | Article |
| id | doaj-art-50b07d52ad234158bbe199ccb791dbbf |
| institution | Kabale University |
| issn | 0013-5194 1350-911X |
| language | English |
| publishDate | 2017-03-01 |
| publisher | Wiley |
| record_format | Article |
| series | Electronics Letters |
| spelling | doaj-art-50b07d52ad234158bbe199ccb791dbbf2025-02-05T12:30:42ZengWileyElectronics Letters0013-51941350-911X2017-03-0153531631810.1049/el.2016.4079Kernelised reference‐wise metric learningMeng Wu0Kai Luo1Daijin Li2Jun Zhou3School of Marine Science and TechnologyNorthwestern Polytechnical UniversityXi'an710072People's Republic of ChinaSchool of Marine Science and TechnologyNorthwestern Polytechnical UniversityXi'an710072People's Republic of ChinaSchool of Marine Science and TechnologyNorthwestern Polytechnical UniversityXi'an710072People's Republic of ChinaDepartment of Electronic EngineeringShanghai Jiao Tong UniversityShanghai200240People's Republic of ChinaUnlike the doublet or triplet constraints, a novel kernelised reference‐wise metric learning is proposed by constructing reference‐wise constraints, which contain similarity information of each sample to all reference samples. After selecting several representative training samples as the reference set, the training data are first mapped into a projected space by the reference sample matrix. Then the problem of metric learning is casted as a multi‐label classification problem under the reference‐wise constraints, in which an l2‐norm regularised least squares canonical correlation analysis can be used. Besides, the formulation is generalised to kernelised version for further boosting the performance. Experiments on two benchmark person re‐identification datasets demonstrate that the approach clearly outperforms the state‐of‐the‐art methods.https://doi.org/10.1049/el.2016.4079Kernelised reference wise metric learningrepresentative training samplesreference sample matrixmultilabel classification problemkernelised version |
| spellingShingle | Meng Wu Kai Luo Daijin Li Jun Zhou Kernelised reference‐wise metric learning Electronics Letters Kernelised reference wise metric learning representative training samples reference sample matrix multilabel classification problem kernelised version |
| title | Kernelised reference‐wise metric learning |
| title_full | Kernelised reference‐wise metric learning |
| title_fullStr | Kernelised reference‐wise metric learning |
| title_full_unstemmed | Kernelised reference‐wise metric learning |
| title_short | Kernelised reference‐wise metric learning |
| title_sort | kernelised reference wise metric learning |
| topic | Kernelised reference wise metric learning representative training samples reference sample matrix multilabel classification problem kernelised version |
| url | https://doi.org/10.1049/el.2016.4079 |
| work_keys_str_mv | AT mengwu kernelisedreferencewisemetriclearning AT kailuo kernelisedreferencewisemetriclearning AT daijinli kernelisedreferencewisemetriclearning AT junzhou kernelisedreferencewisemetriclearning |