Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning
Deep Metric Learning (DML) aims to find representations suitable for zero-shot transfer to a priori unknown test distributions. However, common evaluation protocols only test a single, fixed data split in which train and test classes are assigned randomly. More realistic evaluations should consider...
Main Authors: | , , , , , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | unknown |
Published: |
arXiv
2021
|
Subjects: | |
Online Access: | https://dx.doi.org/10.48550/arxiv.2107.09562 https://arxiv.org/abs/2107.09562 |
id |
ftdatacite:10.48550/arxiv.2107.09562 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.48550/arxiv.2107.09562 2023-05-15T16:01:20+02:00 Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning Milbich, Timo Roth, Karsten Sinha, Samarth Schmidt, Ludwig Ghassemi, Marzyeh Ommer, Björn 2021 https://dx.doi.org/10.48550/arxiv.2107.09562 https://arxiv.org/abs/2107.09562 unknown arXiv arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ Machine Learning cs.LG Computer Vision and Pattern Recognition cs.CV FOS Computer and information sciences Article CreativeWork article Preprint 2021 ftdatacite https://doi.org/10.48550/arxiv.2107.09562 2022-03-10T14:17:29Z Deep Metric Learning (DML) aims to find representations suitable for zero-shot transfer to a priori unknown test distributions. However, common evaluation protocols only test a single, fixed data split in which train and test classes are assigned randomly. More realistic evaluations should consider a broad spectrum of distribution shifts with potentially varying degree and difficulty. In this work, we systematically construct train-test splits of increasing difficulty and present the ooDML benchmark to characterize generalization under out-of-distribution shifts in DML. ooDML is designed to probe the generalization performance on much more challenging, diverse train-to-test distribution shifts. Based on our new benchmark, we conduct a thorough empirical analysis of state-of-the-art DML methods. We find that while generalization tends to consistently degrade with difficulty, some methods are better at retaining performance as the distribution shift increases. Finally, we propose few-shot DML as an efficient way to consistently improve generalization in response to unknown test shifts presented in ooDML. Code available here: https://github.com/CompVis/Characterizing_Generalization_in_DML. : 35th Conference on Neural Information Processing Systems (NeurIPS 2021) 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 |
Machine Learning cs.LG Computer Vision and Pattern Recognition cs.CV FOS Computer and information sciences |
spellingShingle |
Machine Learning cs.LG Computer Vision and Pattern Recognition cs.CV FOS Computer and information sciences Milbich, Timo Roth, Karsten Sinha, Samarth Schmidt, Ludwig Ghassemi, Marzyeh Ommer, Björn Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning |
topic_facet |
Machine Learning cs.LG Computer Vision and Pattern Recognition cs.CV FOS Computer and information sciences |
description |
Deep Metric Learning (DML) aims to find representations suitable for zero-shot transfer to a priori unknown test distributions. However, common evaluation protocols only test a single, fixed data split in which train and test classes are assigned randomly. More realistic evaluations should consider a broad spectrum of distribution shifts with potentially varying degree and difficulty. In this work, we systematically construct train-test splits of increasing difficulty and present the ooDML benchmark to characterize generalization under out-of-distribution shifts in DML. ooDML is designed to probe the generalization performance on much more challenging, diverse train-to-test distribution shifts. Based on our new benchmark, we conduct a thorough empirical analysis of state-of-the-art DML methods. We find that while generalization tends to consistently degrade with difficulty, some methods are better at retaining performance as the distribution shift increases. Finally, we propose few-shot DML as an efficient way to consistently improve generalization in response to unknown test shifts presented in ooDML. Code available here: https://github.com/CompVis/Characterizing_Generalization_in_DML. : 35th Conference on Neural Information Processing Systems (NeurIPS 2021) |
format |
Article in Journal/Newspaper |
author |
Milbich, Timo Roth, Karsten Sinha, Samarth Schmidt, Ludwig Ghassemi, Marzyeh Ommer, Björn |
author_facet |
Milbich, Timo Roth, Karsten Sinha, Samarth Schmidt, Ludwig Ghassemi, Marzyeh Ommer, Björn |
author_sort |
Milbich, Timo |
title |
Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning |
title_short |
Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning |
title_full |
Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning |
title_fullStr |
Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning |
title_full_unstemmed |
Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning |
title_sort |
characterizing generalization under out-of-distribution shifts in deep metric learning |
publisher |
arXiv |
publishDate |
2021 |
url |
https://dx.doi.org/10.48550/arxiv.2107.09562 https://arxiv.org/abs/2107.09562 |
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.2107.09562 |
_version_ |
1766397239283941376 |