Cargando…

Implementation-Independent Representation for Deep Convolutional Neural Networks and Humans in Processing Faces

Deep convolutional neural networks (DCNN) nowadays can match human performance in challenging complex tasks, but it remains unknown whether DCNNs achieve human-like performance through human-like processes. Here we applied a reverse-correlation method to make explicit representations of DCNNs and hu...

Descripción completa

Detalles Bibliográficos
Autores principales: Song, Yiying, Qu, Yukun, Xu, Shan, Liu, Jia
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Frontiers Media S.A. 2021
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7870475/
https://www.ncbi.nlm.nih.gov/pubmed/33574746
http://dx.doi.org/10.3389/fncom.2020.601314
_version_ 1783648824808439808
author Song, Yiying
Qu, Yukun
Xu, Shan
Liu, Jia
author_facet Song, Yiying
Qu, Yukun
Xu, Shan
Liu, Jia
author_sort Song, Yiying
collection PubMed
description Deep convolutional neural networks (DCNN) nowadays can match human performance in challenging complex tasks, but it remains unknown whether DCNNs achieve human-like performance through human-like processes. Here we applied a reverse-correlation method to make explicit representations of DCNNs and humans when performing face gender classification. We found that humans and a typical DCNN, VGG-Face, used similar critical information for this task, which mainly resided at low spatial frequencies. Importantly, the prior task experience, which the VGG-Face was pre-trained to process faces at the subordinate level (i.e., identification) as humans do, seemed necessary for such representational similarity, because AlexNet, a DCNN pre-trained to process objects at the basic level (i.e., categorization), succeeded in gender classification but relied on a completely different representation. In sum, although DCNNs and humans rely on different sets of hardware to process faces, they can use a similar and implementation-independent representation to achieve the same computation goal.
format Online
Article
Text
id pubmed-7870475
institution National Center for Biotechnology Information
language English
publishDate 2021
publisher Frontiers Media S.A.
record_format MEDLINE/PubMed
spelling pubmed-78704752021-02-10 Implementation-Independent Representation for Deep Convolutional Neural Networks and Humans in Processing Faces Song, Yiying Qu, Yukun Xu, Shan Liu, Jia Front Comput Neurosci Neuroscience Deep convolutional neural networks (DCNN) nowadays can match human performance in challenging complex tasks, but it remains unknown whether DCNNs achieve human-like performance through human-like processes. Here we applied a reverse-correlation method to make explicit representations of DCNNs and humans when performing face gender classification. We found that humans and a typical DCNN, VGG-Face, used similar critical information for this task, which mainly resided at low spatial frequencies. Importantly, the prior task experience, which the VGG-Face was pre-trained to process faces at the subordinate level (i.e., identification) as humans do, seemed necessary for such representational similarity, because AlexNet, a DCNN pre-trained to process objects at the basic level (i.e., categorization), succeeded in gender classification but relied on a completely different representation. In sum, although DCNNs and humans rely on different sets of hardware to process faces, they can use a similar and implementation-independent representation to achieve the same computation goal. Frontiers Media S.A. 2021-01-26 /pmc/articles/PMC7870475/ /pubmed/33574746 http://dx.doi.org/10.3389/fncom.2020.601314 Text en Copyright © 2021 Song, Qu, Xu and Liu. http://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
spellingShingle Neuroscience
Song, Yiying
Qu, Yukun
Xu, Shan
Liu, Jia
Implementation-Independent Representation for Deep Convolutional Neural Networks and Humans in Processing Faces
title Implementation-Independent Representation for Deep Convolutional Neural Networks and Humans in Processing Faces
title_full Implementation-Independent Representation for Deep Convolutional Neural Networks and Humans in Processing Faces
title_fullStr Implementation-Independent Representation for Deep Convolutional Neural Networks and Humans in Processing Faces
title_full_unstemmed Implementation-Independent Representation for Deep Convolutional Neural Networks and Humans in Processing Faces
title_short Implementation-Independent Representation for Deep Convolutional Neural Networks and Humans in Processing Faces
title_sort implementation-independent representation for deep convolutional neural networks and humans in processing faces
topic Neuroscience
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7870475/
https://www.ncbi.nlm.nih.gov/pubmed/33574746
http://dx.doi.org/10.3389/fncom.2020.601314
work_keys_str_mv AT songyiying implementationindependentrepresentationfordeepconvolutionalneuralnetworksandhumansinprocessingfaces
AT quyukun implementationindependentrepresentationfordeepconvolutionalneuralnetworksandhumansinprocessingfaces
AT xushan implementationindependentrepresentationfordeepconvolutionalneuralnetworksandhumansinprocessingfaces
AT liujia implementationindependentrepresentationfordeepconvolutionalneuralnetworksandhumansinprocessingfaces