Probabilistic Principal Geodesic Deep Metric Learning
Similarity learning which is useful for the purpose of comparing various characteristics of images in the computer vision field has been often applied for deep metric learning (DML). Also, a lot of combinations of pairwise similarity metrics such as Euclidean distance and cosine similarity have been...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9681811/ |
| _version_ | 1818504223894011904 |
|---|---|
| author | Dae Ha Kim Byung Cheol Song |
| author_facet | Dae Ha Kim Byung Cheol Song |
| author_sort | Dae Ha Kim |
| collection | DOAJ |
| description | Similarity learning which is useful for the purpose of comparing various characteristics of images in the computer vision field has been often applied for deep metric learning (DML). Also, a lot of combinations of pairwise similarity metrics such as Euclidean distance and cosine similarity have been studied actively. However, such a local similarity-based approach can be rather a bottleneck for a retrieval task in which global characteristics of images must be considered important. Therefore, this paper proposes a new similarity metric structure that considers the local similarity as well as the global characteristic on the representation space, i.e., class variability. Also, based on an insight that better class variability analysis can be accomplished on the Stiefel (or Riemannian) manifold, manifold geometry is employed to generate class variability information. Finally, we show that the proposed method designed through in-depth analysis of generalization bound of DML outperforms conventional DML methods theoretically and experimentally. |
| first_indexed | 2024-12-10T21:34:18Z |
| format | Article |
| id | doaj.art-88bc23efd1e640f2affb6f0acaa11c37 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-10T21:34:18Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-88bc23efd1e640f2affb6f0acaa11c372022-12-22T01:32:41ZengIEEEIEEE Access2169-35362022-01-01107439745910.1109/ACCESS.2022.31431299681811Probabilistic Principal Geodesic Deep Metric LearningDae Ha Kim0https://orcid.org/0000-0003-3838-126XByung Cheol Song1https://orcid.org/0000-0001-8742-3433Department of Electrical and Computer Engineering, Inha University, Incheon, South KoreaDepartment of Electrical and Computer Engineering, Inha University, Incheon, South KoreaSimilarity learning which is useful for the purpose of comparing various characteristics of images in the computer vision field has been often applied for deep metric learning (DML). Also, a lot of combinations of pairwise similarity metrics such as Euclidean distance and cosine similarity have been studied actively. However, such a local similarity-based approach can be rather a bottleneck for a retrieval task in which global characteristics of images must be considered important. Therefore, this paper proposes a new similarity metric structure that considers the local similarity as well as the global characteristic on the representation space, i.e., class variability. Also, based on an insight that better class variability analysis can be accomplished on the Stiefel (or Riemannian) manifold, manifold geometry is employed to generate class variability information. Finally, we show that the proposed method designed through in-depth analysis of generalization bound of DML outperforms conventional DML methods theoretically and experimentally.https://ieeexplore.ieee.org/document/9681811/Deep metric learningimage retrievalStiefel manifoldnon-linear mapping |
| spellingShingle | Dae Ha Kim Byung Cheol Song Probabilistic Principal Geodesic Deep Metric Learning IEEE Access Deep metric learning image retrieval Stiefel manifold non-linear mapping |
| title | Probabilistic Principal Geodesic Deep Metric Learning |
| title_full | Probabilistic Principal Geodesic Deep Metric Learning |
| title_fullStr | Probabilistic Principal Geodesic Deep Metric Learning |
| title_full_unstemmed | Probabilistic Principal Geodesic Deep Metric Learning |
| title_short | Probabilistic Principal Geodesic Deep Metric Learning |
| title_sort | probabilistic principal geodesic deep metric learning |
| topic | Deep metric learning image retrieval Stiefel manifold non-linear mapping |
| url | https://ieeexplore.ieee.org/document/9681811/ |
| work_keys_str_mv | AT daehakim probabilisticprincipalgeodesicdeepmetriclearning AT byungcheolsong probabilisticprincipalgeodesicdeepmetriclearning |