1903 - Making Every Label Count: Handling Semantic Imprecision By Integrating Domain Knowledge
ICPR Browser Link: https://ailb-web.ing.unimore.it/icpr/paper/287/nn Abstract: Noisy data, crawled from the web or supplied by volunteers such as Mechanical Turkers or citizen scientists, is considered an alternative to professionally labeled data. There has been research focused on mitigating the e...
Main Authors: | , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | unknown |
Published: |
Underline Science Inc.
2020
|
Subjects: | |
Online Access: | https://dx.doi.org/10.48448/gz4x-8r37 https://underline.io/lecture/11702-1903---making-every-label-count-handling-semantic-imprecision-by-integrating-domain-knowledge |
id |
ftdatacite:10.48448/gz4x-8r37 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.48448/gz4x-8r37 2023-05-15T18:20:04+02:00 1903 - Making Every Label Count: Handling Semantic Imprecision By Integrating Domain Knowledge 25th International Conference on Pattern Recognition 2021 Brust, Clemens-Alexander 2020 https://dx.doi.org/10.48448/gz4x-8r37 https://underline.io/lecture/11702-1903---making-every-label-count-handling-semantic-imprecision-by-integrating-domain-knowledge unknown Underline Science Inc. Artificial Intelligence Machine Learning Neural Network MediaObject article Conference talk Audiovisual 2020 ftdatacite https://doi.org/10.48448/gz4x-8r37 2022-02-09T11:27:25Z ICPR Browser Link: https://ailb-web.ing.unimore.it/icpr/paper/287/nn Abstract: Noisy data, crawled from the web or supplied by volunteers such as Mechanical Turkers or citizen scientists, is considered an alternative to professionally labeled data. There has been research focused on mitigating the effects of label noise. It is typically modeled as inaccuracy, where the correct label is replaced by an incorrect label from the same set. We consider an additional dimension of label noise: imprecision. For example, a non-breeding snow bunting is labeled as a bird. This label is correct, but not as precise as the task requires. Standard softmax classifiers cannot learn from such a weak label because they consider all classes mutually exclusive, which non-breeding snow bunting and bird are not. We propose CHILLAX (Class Hierarchies for Imprecise Label Learning and Annotation eXtrapolation), a method based on hierarchical classification, to fully utilize labels of any precision. Experiments on noisy variants of NABirds and ILSVRC2012 show that our method outperforms strong baselines by as much as 16.4 percentage points, and the current state of the art by up to 3.9 percentage points. 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 |
Artificial Intelligence Machine Learning Neural Network |
spellingShingle |
Artificial Intelligence Machine Learning Neural Network 25th International Conference on Pattern Recognition 2021 Brust, Clemens-Alexander 1903 - Making Every Label Count: Handling Semantic Imprecision By Integrating Domain Knowledge |
topic_facet |
Artificial Intelligence Machine Learning Neural Network |
description |
ICPR Browser Link: https://ailb-web.ing.unimore.it/icpr/paper/287/nn Abstract: Noisy data, crawled from the web or supplied by volunteers such as Mechanical Turkers or citizen scientists, is considered an alternative to professionally labeled data. There has been research focused on mitigating the effects of label noise. It is typically modeled as inaccuracy, where the correct label is replaced by an incorrect label from the same set. We consider an additional dimension of label noise: imprecision. For example, a non-breeding snow bunting is labeled as a bird. This label is correct, but not as precise as the task requires. Standard softmax classifiers cannot learn from such a weak label because they consider all classes mutually exclusive, which non-breeding snow bunting and bird are not. We propose CHILLAX (Class Hierarchies for Imprecise Label Learning and Annotation eXtrapolation), a method based on hierarchical classification, to fully utilize labels of any precision. Experiments on noisy variants of NABirds and ILSVRC2012 show that our method outperforms strong baselines by as much as 16.4 percentage points, and the current state of the art by up to 3.9 percentage points. |
format |
Article in Journal/Newspaper |
author |
25th International Conference on Pattern Recognition 2021 Brust, Clemens-Alexander |
author_facet |
25th International Conference on Pattern Recognition 2021 Brust, Clemens-Alexander |
author_sort |
25th International Conference on Pattern Recognition 2021 |
title |
1903 - Making Every Label Count: Handling Semantic Imprecision By Integrating Domain Knowledge |
title_short |
1903 - Making Every Label Count: Handling Semantic Imprecision By Integrating Domain Knowledge |
title_full |
1903 - Making Every Label Count: Handling Semantic Imprecision By Integrating Domain Knowledge |
title_fullStr |
1903 - Making Every Label Count: Handling Semantic Imprecision By Integrating Domain Knowledge |
title_full_unstemmed |
1903 - Making Every Label Count: Handling Semantic Imprecision By Integrating Domain Knowledge |
title_sort |
1903 - making every label count: handling semantic imprecision by integrating domain knowledge |
publisher |
Underline Science Inc. |
publishDate |
2020 |
url |
https://dx.doi.org/10.48448/gz4x-8r37 https://underline.io/lecture/11702-1903---making-every-label-count-handling-semantic-imprecision-by-integrating-domain-knowledge |
genre |
Snow Bunting |
genre_facet |
Snow Bunting |
op_doi |
https://doi.org/10.48448/gz4x-8r37 |
_version_ |
1766197540174168064 |