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

被引:7
|
作者
Song, Yiying [1 ]
Qu, Yukun [2 ]
Xu, Shan [1 ]
Liu, Jia [3 ,4 ]
机构
[1] Beijing Normal Univ, Fac Psychol, Beijing Key Lab Appl Expt Psychol, Beijing, Peoples R China
[2] Beijing Normal Univ, State Key Lab Cognit Neurosci & Learning, Beijing, Peoples R China
[3] Tsinghua Univ, Dept Psychol, Beijing, Peoples R China
[4] Tsinghua Univ, Tsinghua Lab Brain & Intelligence, Beijing, Peoples R China
基金
中国国家自然科学基金;
关键词
deep convolutional neural network; face recognition; reverse correlation analysis; face representation; visual intelligence; OBJECT; INFORMATION;
D O I
10.3389/fncom.2020.601314
中图分类号
Q [生物科学];
学科分类号
07 ; 0710 ; 09 ;
摘要
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.
引用
收藏
页数:9
相关论文
共 50 条
  • [41] Time-Independent Prediction of Burn Depth Using Deep Convolutional Neural Networks
    Cirillo, Marco Domenico
    Mirdell, Robin
    Sjoberg, Folke
    Pham, Tuan D.
    JOURNAL OF BURN CARE & RESEARCH, 2019, 40 (06): : 857 - 863
  • [42] Generalisation in humans and deep neural networks
    Geirhos, Robert
    Temme, Carlos R. Medina
    Rauber, Jonas
    Schuett, Heiko H.
    Bethge, Matthias
    Wichmann, Felix A.
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 31 (NIPS 2018), 2018, 31
  • [43] FPGA based Flexible Implementation of Light Weight Inference on Deep Convolutional Neural Networks
    Dawwd, Shefa
    INTERNATIONAL ARAB JOURNAL OF INFORMATION TECHNOLOGY, 2024, 21 (03) : 408 - 417
  • [44] Efficient Implementation of Convolutional Neural Networks on FPGA
    Hadnagy, A.
    Feher, B.
    Kovacshazy, T.
    2018 19TH INTERNATIONAL CARPATHIAN CONTROL CONFERENCE (ICCC), 2018, : 359 - 364
  • [45] Convolutional Neural Networks Implementation for Chili Classification
    Purwaningsih, Tuti
    Anjani, Imania Ayu
    Utami, Pertiwi Bekti
    2018 INTERNATIONAL SYMPOSIUM ON ADVANCED INTELLIGENT INFORMATICS (SAIN), 2018, : 190 - 194
  • [46] FPGA Implementation and Acceleration of Convolutional Neural Networks
    Pisharody, Jayanth N.
    Pranav, K. B.
    Ranjitha, M.
    Rajeshwari, B.
    2021 6TH INTERNATIONAL CONFERENCE FOR CONVERGENCE IN TECHNOLOGY (I2CT), 2021,
  • [47] COLOR REPRESENTATION IN DEEP NEURAL NETWORKS
    Engilberge, Martin
    Collins, Edo
    Susstrunk, Sabine
    2017 24TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2017, : 2786 - 2790
  • [48] Noise Convolutional Neural Networks and FPGA Implementation
    Munakala, Atsuki
    Nakahara, IIiroki
    Sato, Shimpei
    2019 IEEE 49TH INTERNATIONAL SYMPOSIUM ON MULTIPLE-VALUED LOGIC (ISMVL), 2019, : 85 - 90
  • [49] The Representation of Speech in Deep Neural Networks
    Scharenborg, Odette
    van der Gouw, Nikki
    Larson, Martha
    Marchiori, Elena
    MULTIMEDIA MODELING, MMM 2019, PT II, 2019, 11296 : 194 - 205
  • [50] Comparing Object Recognition in Humans and Deep Convolutional Neural Networks-An Eye Tracking Study
    van Dyck, Leonard Elia
    Kwitt, Roland
    Denzler, Sebastian Jochen
    Gruber, Walter Roland
    FRONTIERS IN NEUROSCIENCE, 2021, 15