Self-Supervised Learning from Semantically Imprecise Data
Learning from imprecise labels such as animal or bird, but making precise predictions like snow bunting at inference time is an important capability for any classifier when expertly labeled training data is scarce. Contributions by volunteers or results of web crawling lack precision in this manner,...
Published in: | Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications |
---|---|
Main Authors: | , , |
Other Authors: | , , |
Format: | Conference Object |
Language: | English |
Published: |
SCITEPRESS
2022
|
Subjects: | |
Online Access: | https://elib.dlr.de/186359/ https://elib.dlr.de/186359/1/107667.pdf https://www.scitepress.org/PublicationsDetail.aspx?ID=PSP7VmVv1RY=&t=1 |
id |
ftdlr:oai:elib.dlr.de:186359 |
---|---|
record_format |
openpolar |
spelling |
ftdlr:oai:elib.dlr.de:186359 2024-05-19T07:48:27+00:00 Self-Supervised Learning from Semantically Imprecise Data Brust, Clemens-Alexander Barz, Björn Denzler, Joachim Farinella, Giovanni Maria Radeva, Petia Bouatouch, Kadi 2022 application/pdf https://elib.dlr.de/186359/ https://elib.dlr.de/186359/1/107667.pdf https://www.scitepress.org/PublicationsDetail.aspx?ID=PSP7VmVv1RY=&t=1 en eng SCITEPRESS https://elib.dlr.de/186359/1/107667.pdf Brust, Clemens-Alexander und Barz, Björn und Denzler, Joachim (2022) Self-Supervised Learning from Semantically Imprecise Data. In: 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2022, 5, Seiten 27-35. SCITEPRESS. Computer Vision Theory and Applications (VISAPP), 2022-02-06 - 2022-02-08, Online. doi:10.5220/0010766700003124 <https://doi.org/10.5220/0010766700003124>. ISBN 978-989-758-555-5. ISSN 2184-4321. cc_by_nc_nd Datenanalyse und -intelligenz Konferenzbeitrag PeerReviewed 2022 ftdlr https://doi.org/10.5220/0010766700003124 2024-04-25T00:59:55Z Learning from imprecise labels such as animal or bird, but making precise predictions like snow bunting at inference time is an important capability for any classifier when expertly labeled training data is scarce. Contributions by volunteers or results of web crawling lack precision in this manner, but are still valuable. And crucially, these weakly labeled examples are available in larger quantities for lower cost than high-quality bespoke training data. CHILLAX, a recently proposed method to tackle this task, leverages a hierarchical classifier to learn from imprecise labels. However, it has two major limitations. First, it does not learn from examples labeled as the root of the hierarchy, e.g., object. Second, an extrapolation of annotations to precise labels is only performed at test time, where confident extrapolations could be already used as training data. In this work, we extend CHILLAX with a self-supervised scheme using constrained semantic extrapolation to generate pseudo-labels. This addresses the second concern, which in turn solves the first problem, enabling an even weaker supervision requirement than CHILLAX. We evaluate our approach empirically, showing that our method allows for a consistent accuracy improvement of 0.84 to 1.19 percent points over CHILLAX and is suitable as a drop-in replacement without any negative consequences such as longer training times. Conference Object Snow Bunting German Aerospace Center: elib - DLR electronic library Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications 27 35 |
institution |
Open Polar |
collection |
German Aerospace Center: elib - DLR electronic library |
op_collection_id |
ftdlr |
language |
English |
topic |
Datenanalyse und -intelligenz |
spellingShingle |
Datenanalyse und -intelligenz Brust, Clemens-Alexander Barz, Björn Denzler, Joachim Self-Supervised Learning from Semantically Imprecise Data |
topic_facet |
Datenanalyse und -intelligenz |
description |
Learning from imprecise labels such as animal or bird, but making precise predictions like snow bunting at inference time is an important capability for any classifier when expertly labeled training data is scarce. Contributions by volunteers or results of web crawling lack precision in this manner, but are still valuable. And crucially, these weakly labeled examples are available in larger quantities for lower cost than high-quality bespoke training data. CHILLAX, a recently proposed method to tackle this task, leverages a hierarchical classifier to learn from imprecise labels. However, it has two major limitations. First, it does not learn from examples labeled as the root of the hierarchy, e.g., object. Second, an extrapolation of annotations to precise labels is only performed at test time, where confident extrapolations could be already used as training data. In this work, we extend CHILLAX with a self-supervised scheme using constrained semantic extrapolation to generate pseudo-labels. This addresses the second concern, which in turn solves the first problem, enabling an even weaker supervision requirement than CHILLAX. We evaluate our approach empirically, showing that our method allows for a consistent accuracy improvement of 0.84 to 1.19 percent points over CHILLAX and is suitable as a drop-in replacement without any negative consequences such as longer training times. |
author2 |
Farinella, Giovanni Maria Radeva, Petia Bouatouch, Kadi |
format |
Conference Object |
author |
Brust, Clemens-Alexander Barz, Björn Denzler, Joachim |
author_facet |
Brust, Clemens-Alexander Barz, Björn Denzler, Joachim |
author_sort |
Brust, Clemens-Alexander |
title |
Self-Supervised Learning from Semantically Imprecise Data |
title_short |
Self-Supervised Learning from Semantically Imprecise Data |
title_full |
Self-Supervised Learning from Semantically Imprecise Data |
title_fullStr |
Self-Supervised Learning from Semantically Imprecise Data |
title_full_unstemmed |
Self-Supervised Learning from Semantically Imprecise Data |
title_sort |
self-supervised learning from semantically imprecise data |
publisher |
SCITEPRESS |
publishDate |
2022 |
url |
https://elib.dlr.de/186359/ https://elib.dlr.de/186359/1/107667.pdf https://www.scitepress.org/PublicationsDetail.aspx?ID=PSP7VmVv1RY=&t=1 |
genre |
Snow Bunting |
genre_facet |
Snow Bunting |
op_relation |
https://elib.dlr.de/186359/1/107667.pdf Brust, Clemens-Alexander und Barz, Björn und Denzler, Joachim (2022) Self-Supervised Learning from Semantically Imprecise Data. In: 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2022, 5, Seiten 27-35. SCITEPRESS. Computer Vision Theory and Applications (VISAPP), 2022-02-06 - 2022-02-08, Online. doi:10.5220/0010766700003124 <https://doi.org/10.5220/0010766700003124>. ISBN 978-989-758-555-5. ISSN 2184-4321. |
op_rights |
cc_by_nc_nd |
op_doi |
https://doi.org/10.5220/0010766700003124 |
container_title |
Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications |
container_start_page |
27 |
op_container_end_page |
35 |
_version_ |
1799466710711926784 |