Cargando…
A dataset for evaluating one-shot categorization of novel object classes
With the advent of deep convolutional neural networks, machines now rival humans in terms of object categorization. The neural networks solve categorization with a hierarchical organization that shares a striking resemblance to their biological counterpart, leading to their status as a standard mode...
Autores principales: | , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Elsevier
2020
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7044642/ https://www.ncbi.nlm.nih.gov/pubmed/32140517 http://dx.doi.org/10.1016/j.dib.2020.105302 |
_version_ | 1783501617801199616 |
---|---|
author | Morgenstern, Yaniv Schmidt, Filipp Fleming, Roland W. |
author_facet | Morgenstern, Yaniv Schmidt, Filipp Fleming, Roland W. |
author_sort | Morgenstern, Yaniv |
collection | PubMed |
description | With the advent of deep convolutional neural networks, machines now rival humans in terms of object categorization. The neural networks solve categorization with a hierarchical organization that shares a striking resemblance to their biological counterpart, leading to their status as a standard model of object recognition in biological vision. Despite training on thousands of images of object categories, however, machine-learning networks are poorer generalizers, often fooled by adversarial images with very simple image manipulations that humans easily distinguish as a false image. Humans, on the other hand, can generalize object classes from very few samples. Here we provide a dataset of novel object classifications in humans. We gathered thousands of crowd-sourced human responses to novel objects embedded either with 1 or 16 context sample(s). Human decisions and stimuli together have the potential to be re-used (1) as a tool to better understand the nature of the gap in category learning from few samples between human and machine, and (2) as a benchmark of generalization across machine learning networks. |
format | Online Article Text |
id | pubmed-7044642 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2020 |
publisher | Elsevier |
record_format | MEDLINE/PubMed |
spelling | pubmed-70446422020-03-05 A dataset for evaluating one-shot categorization of novel object classes Morgenstern, Yaniv Schmidt, Filipp Fleming, Roland W. Data Brief Computer Science With the advent of deep convolutional neural networks, machines now rival humans in terms of object categorization. The neural networks solve categorization with a hierarchical organization that shares a striking resemblance to their biological counterpart, leading to their status as a standard model of object recognition in biological vision. Despite training on thousands of images of object categories, however, machine-learning networks are poorer generalizers, often fooled by adversarial images with very simple image manipulations that humans easily distinguish as a false image. Humans, on the other hand, can generalize object classes from very few samples. Here we provide a dataset of novel object classifications in humans. We gathered thousands of crowd-sourced human responses to novel objects embedded either with 1 or 16 context sample(s). Human decisions and stimuli together have the potential to be re-used (1) as a tool to better understand the nature of the gap in category learning from few samples between human and machine, and (2) as a benchmark of generalization across machine learning networks. Elsevier 2020-02-21 /pmc/articles/PMC7044642/ /pubmed/32140517 http://dx.doi.org/10.1016/j.dib.2020.105302 Text en © 2020 The Author(s) http://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). |
spellingShingle | Computer Science Morgenstern, Yaniv Schmidt, Filipp Fleming, Roland W. A dataset for evaluating one-shot categorization of novel object classes |
title | A dataset for evaluating one-shot categorization of novel object classes |
title_full | A dataset for evaluating one-shot categorization of novel object classes |
title_fullStr | A dataset for evaluating one-shot categorization of novel object classes |
title_full_unstemmed | A dataset for evaluating one-shot categorization of novel object classes |
title_short | A dataset for evaluating one-shot categorization of novel object classes |
title_sort | dataset for evaluating one-shot categorization of novel object classes |
topic | Computer Science |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7044642/ https://www.ncbi.nlm.nih.gov/pubmed/32140517 http://dx.doi.org/10.1016/j.dib.2020.105302 |
work_keys_str_mv | AT morgensternyaniv adatasetforevaluatingoneshotcategorizationofnovelobjectclasses AT schmidtfilipp adatasetforevaluatingoneshotcategorizationofnovelobjectclasses AT flemingrolandw adatasetforevaluatingoneshotcategorizationofnovelobjectclasses AT morgensternyaniv datasetforevaluatingoneshotcategorizationofnovelobjectclasses AT schmidtfilipp datasetforevaluatingoneshotcategorizationofnovelobjectclasses AT flemingrolandw datasetforevaluatingoneshotcategorizationofnovelobjectclasses |