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 cra...

Full description

Bibliographic Details
Main Authors: Brust, Clemens-Alexander, Barz, Björn, Denzler, Joachim
Format: Article in Journal/Newspaper
Language:unknown
Published: arXiv 2021
Subjects:
Online Access:https://dx.doi.org/10.48550/arxiv.2104.10901
https://arxiv.org/abs/2104.10901
id ftdatacite:10.48550/arxiv.2104.10901
record_format openpolar
spelling ftdatacite:10.48550/arxiv.2104.10901 2023-05-15T18:20:04+02:00 Self-Supervised Learning from Semantically Imprecise Data Brust, Clemens-Alexander Barz, Björn Denzler, Joachim 2021 https://dx.doi.org/10.48550/arxiv.2104.10901 https://arxiv.org/abs/2104.10901 unknown arXiv arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ Computer Vision and Pattern Recognition cs.CV FOS Computer and information sciences Article CreativeWork article Preprint 2021 ftdatacite https://doi.org/10.48550/arxiv.2104.10901 2022-03-10T14:36:28Z 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. : 9 pages. Accepted for publication at VISAPP 2022 Article in Journal/Newspaper Snow Bunting 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
FOS Computer and information sciences
spellingShingle Computer Vision and Pattern Recognition cs.CV
FOS Computer and information sciences
Brust, Clemens-Alexander
Barz, Björn
Denzler, Joachim
Self-Supervised Learning from Semantically Imprecise Data
topic_facet Computer Vision and Pattern Recognition cs.CV
FOS Computer and information sciences
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. : 9 pages. Accepted for publication at VISAPP 2022
format Article in Journal/Newspaper
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 arXiv
publishDate 2021
url https://dx.doi.org/10.48550/arxiv.2104.10901
https://arxiv.org/abs/2104.10901
genre Snow Bunting
genre_facet Snow Bunting
op_rights arXiv.org perpetual, non-exclusive license
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
op_doi https://doi.org/10.48550/arxiv.2104.10901
_version_ 1766197542620495872