Cargando…

Deep Networks Can Resemble Human Feed-forward Vision in Invariant Object Recognition

Deep convolutional neural networks (DCNNs) have attracted much attention recently, and have shown to be able to recognize thousands of object categories in natural image databases. Their architecture is somewhat similar to that of the human visual system: both use restricted receptive fields, and a...

Descripción completa

Detalles Bibliográficos
Autores principales: Kheradpisheh, Saeed Reza, Ghodrati, Masoud, Ganjtabesh, Mohammad, Masquelier, Timothée
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Nature Publishing Group 2016
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5013454/
https://www.ncbi.nlm.nih.gov/pubmed/27601096
http://dx.doi.org/10.1038/srep32672
_version_ 1782452168829173760
author Kheradpisheh, Saeed Reza
Ghodrati, Masoud
Ganjtabesh, Mohammad
Masquelier, Timothée
author_facet Kheradpisheh, Saeed Reza
Ghodrati, Masoud
Ganjtabesh, Mohammad
Masquelier, Timothée
author_sort Kheradpisheh, Saeed Reza
collection PubMed
description Deep convolutional neural networks (DCNNs) have attracted much attention recently, and have shown to be able to recognize thousands of object categories in natural image databases. Their architecture is somewhat similar to that of the human visual system: both use restricted receptive fields, and a hierarchy of layers which progressively extract more and more abstracted features. Yet it is unknown whether DCNNs match human performance at the task of view-invariant object recognition, whether they make similar errors and use similar representations for this task, and whether the answers depend on the magnitude of the viewpoint variations. To investigate these issues, we benchmarked eight state-of-the-art DCNNs, the HMAX model, and a baseline shallow model and compared their results to those of humans with backward masking. Unlike in all previous DCNN studies, we carefully controlled the magnitude of the viewpoint variations to demonstrate that shallow nets can outperform deep nets and humans when variations are weak. When facing larger variations, however, more layers were needed to match human performance and error distributions, and to have representations that are consistent with human behavior. A very deep net with 18 layers even outperformed humans at the highest variation level, using the most human-like representations.
format Online
Article
Text
id pubmed-5013454
institution National Center for Biotechnology Information
language English
publishDate 2016
publisher Nature Publishing Group
record_format MEDLINE/PubMed
spelling pubmed-50134542016-09-12 Deep Networks Can Resemble Human Feed-forward Vision in Invariant Object Recognition Kheradpisheh, Saeed Reza Ghodrati, Masoud Ganjtabesh, Mohammad Masquelier, Timothée Sci Rep Article Deep convolutional neural networks (DCNNs) have attracted much attention recently, and have shown to be able to recognize thousands of object categories in natural image databases. Their architecture is somewhat similar to that of the human visual system: both use restricted receptive fields, and a hierarchy of layers which progressively extract more and more abstracted features. Yet it is unknown whether DCNNs match human performance at the task of view-invariant object recognition, whether they make similar errors and use similar representations for this task, and whether the answers depend on the magnitude of the viewpoint variations. To investigate these issues, we benchmarked eight state-of-the-art DCNNs, the HMAX model, and a baseline shallow model and compared their results to those of humans with backward masking. Unlike in all previous DCNN studies, we carefully controlled the magnitude of the viewpoint variations to demonstrate that shallow nets can outperform deep nets and humans when variations are weak. When facing larger variations, however, more layers were needed to match human performance and error distributions, and to have representations that are consistent with human behavior. A very deep net with 18 layers even outperformed humans at the highest variation level, using the most human-like representations. Nature Publishing Group 2016-09-07 /pmc/articles/PMC5013454/ /pubmed/27601096 http://dx.doi.org/10.1038/srep32672 Text en Copyright © 2016, The Author(s) http://creativecommons.org/licenses/by/4.0/ This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
spellingShingle Article
Kheradpisheh, Saeed Reza
Ghodrati, Masoud
Ganjtabesh, Mohammad
Masquelier, Timothée
Deep Networks Can Resemble Human Feed-forward Vision in Invariant Object Recognition
title Deep Networks Can Resemble Human Feed-forward Vision in Invariant Object Recognition
title_full Deep Networks Can Resemble Human Feed-forward Vision in Invariant Object Recognition
title_fullStr Deep Networks Can Resemble Human Feed-forward Vision in Invariant Object Recognition
title_full_unstemmed Deep Networks Can Resemble Human Feed-forward Vision in Invariant Object Recognition
title_short Deep Networks Can Resemble Human Feed-forward Vision in Invariant Object Recognition
title_sort deep networks can resemble human feed-forward vision in invariant object recognition
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5013454/
https://www.ncbi.nlm.nih.gov/pubmed/27601096
http://dx.doi.org/10.1038/srep32672
work_keys_str_mv AT kheradpishehsaeedreza deepnetworkscanresemblehumanfeedforwardvisionininvariantobjectrecognition
AT ghodratimasoud deepnetworkscanresemblehumanfeedforwardvisionininvariantobjectrecognition
AT ganjtabeshmohammad deepnetworkscanresemblehumanfeedforwardvisionininvariantobjectrecognition
AT masqueliertimothee deepnetworkscanresemblehumanfeedforwardvisionininvariantobjectrecognition