Deep Metric Learning with Chance Constraints ...
Deep metric learning (DML) aims to minimize empirical expected loss of the pairwise intra-/inter- class proximity violations in the embedding space. We relate DML to feasibility problem of finite chance constraints. We show that minimizer of proxy-based DML satisfies certain chance constraints, and...
Main Authors: | , , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | unknown |
Published: |
arXiv
2022
|
Subjects: | |
Online Access: | https://dx.doi.org/10.48550/arxiv.2209.09060 https://arxiv.org/abs/2209.09060 |
id |
ftdatacite:10.48550/arxiv.2209.09060 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.48550/arxiv.2209.09060 2023-11-05T03:41:36+01:00 Deep Metric Learning with Chance Constraints ... Gurbuz, Yeti Z. Can, Ogul Alatan, A. Aydin 2022 https://dx.doi.org/10.48550/arxiv.2209.09060 https://arxiv.org/abs/2209.09060 unknown arXiv arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ Computer Vision and Pattern Recognition cs.CV Machine Learning cs.LG Machine Learning stat.ML FOS Computer and information sciences Article article CreativeWork Preprint 2022 ftdatacite https://doi.org/10.48550/arxiv.2209.09060 2023-10-09T10:43:42Z Deep metric learning (DML) aims to minimize empirical expected loss of the pairwise intra-/inter- class proximity violations in the embedding space. We relate DML to feasibility problem of finite chance constraints. We show that minimizer of proxy-based DML satisfies certain chance constraints, and that the worst case generalization performance of the proxy-based methods can be characterized by the radius of the smallest ball around a class proxy to cover the entire domain of the corresponding class samples, suggesting multiple proxies per class helps performance. To provide a scalable algorithm as well as exploiting more proxies, we consider the chance constraints implied by the minimizers of proxy-based DML instances and reformulate DML as finding a feasible point in intersection of such constraints, resulting in a problem to be approximately solved by iterative projections. Simply put, we repeatedly train a regularized proxy-based loss and re-initialize the proxies with the embeddings of the deliberately ... : Accepted as a conference paper at IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2024 ... Article in Journal/Newspaper DML DataCite Metadata Store (German National Library of Science and Technology) |
institution |
Open Polar |
collection |
DataCite Metadata Store (German National Library of Science and Technology) |
op_collection_id |
ftdatacite |
language |
unknown |
topic |
Computer Vision and Pattern Recognition cs.CV Machine Learning cs.LG Machine Learning stat.ML FOS Computer and information sciences |
spellingShingle |
Computer Vision and Pattern Recognition cs.CV Machine Learning cs.LG Machine Learning stat.ML FOS Computer and information sciences Gurbuz, Yeti Z. Can, Ogul Alatan, A. Aydin Deep Metric Learning with Chance Constraints ... |
topic_facet |
Computer Vision and Pattern Recognition cs.CV Machine Learning cs.LG Machine Learning stat.ML FOS Computer and information sciences |
description |
Deep metric learning (DML) aims to minimize empirical expected loss of the pairwise intra-/inter- class proximity violations in the embedding space. We relate DML to feasibility problem of finite chance constraints. We show that minimizer of proxy-based DML satisfies certain chance constraints, and that the worst case generalization performance of the proxy-based methods can be characterized by the radius of the smallest ball around a class proxy to cover the entire domain of the corresponding class samples, suggesting multiple proxies per class helps performance. To provide a scalable algorithm as well as exploiting more proxies, we consider the chance constraints implied by the minimizers of proxy-based DML instances and reformulate DML as finding a feasible point in intersection of such constraints, resulting in a problem to be approximately solved by iterative projections. Simply put, we repeatedly train a regularized proxy-based loss and re-initialize the proxies with the embeddings of the deliberately ... : Accepted as a conference paper at IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2024 ... |
format |
Article in Journal/Newspaper |
author |
Gurbuz, Yeti Z. Can, Ogul Alatan, A. Aydin |
author_facet |
Gurbuz, Yeti Z. Can, Ogul Alatan, A. Aydin |
author_sort |
Gurbuz, Yeti Z. |
title |
Deep Metric Learning with Chance Constraints ... |
title_short |
Deep Metric Learning with Chance Constraints ... |
title_full |
Deep Metric Learning with Chance Constraints ... |
title_fullStr |
Deep Metric Learning with Chance Constraints ... |
title_full_unstemmed |
Deep Metric Learning with Chance Constraints ... |
title_sort |
deep metric learning with chance constraints ... |
publisher |
arXiv |
publishDate |
2022 |
url |
https://dx.doi.org/10.48550/arxiv.2209.09060 https://arxiv.org/abs/2209.09060 |
genre |
DML |
genre_facet |
DML |
op_rights |
arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
op_doi |
https://doi.org/10.48550/arxiv.2209.09060 |
_version_ |
1781698027394695168 |