Fixing the problems of deep neural networks will require better training data and learning algorithms

被引：0

作者：

Bowers, Jeffrey S. ^{[1
]}

Malhotra, Gaurav ^{[1
]}

Dujmovic, Marin ^{[1
]}

Montero, Milton Llera ^{[1
]}

Tsvetkov, Christian ^{[1
]}

Biscione, Valerio ^{[1
]}

Puebla, Guillermo ^{[1
]}

Adolfi, Federico ^{[1
,2
]}

Hummel, John E. ^{[3
]}

Heaton, Rachel F. ^{[3
]}

Evans, Benjamin D. ^{[4
]}

Mitchell, Jeffrey ^{[4
]}

Blything, Ryan ^{[5
]}

Anderson, Barton L. ^{[6
]}

Storrs, Katherine R. ^{[7
]}

Fleming, Roland W. ^{[8
,9
]}

Bever, Thomas G. ^{[10
]}

Chomsky, Noam ^{[10
]}

Fong, Sandiway ^{[10
]}

Piattelli-Palmarini, Massimo ^{[10
]}

Chandran, Keerthi S. ^{[11
,12
]}

Paul, Amrita Mukherjee ^{[11
,13
]}

Paul, Avijit ^{[14
]}

Ghosh, Kuntal ^{[12
]}

de Vries, Jelmer Philip ^{[15
]}

Flachot, Alban ^{[16
]}

Morimoto, Takuma ^{[15
,17
]}

Gegenfurtner, Karl R. ^{[15
]}

DiCarlo, James J. ^{[18
,19
]}

Yamins, Daniel L. K. ^{[20
]}

Ferguson, Michael E. ^{[18
,19
]}

Fedorenko, Evelina ^{[18
,19
]}

Bethge, Matthias ^{[21
]}

Bonnen, Tyler

Schrimpf, Martin ^{[18
,19
,22
]}

German, Joseph Scott ^{[23
]}

Jacobs, Robert A. ^{[24
]}

Golan, Tal ^{[25
]}

Taylor, JohnMark ^{[26
]}

Schutt, Heiko ^{[26
,27
]}

Peters, Benjamin ^{[28
]}

Sommers, Rowan P. ^{[29
]}

Seeliger, Katja ^{[30
]}

Doerig, Adrien ^{[31
]}

Linton, Paul ^{[32
,33
]}

Konkle, Talia ^{[34
]}

van Gerven, Marcel ^{[35
]}

Kording, Konrad ^{[36
,37
]}

Richards, Blake ^{[38
,39
,40
,41
]}

Kietzmann, Tim C.

机构：

[1] Univ Bristol, Sch Psychol Sci, Bristol, Avon, England

[2] Max Planck Gesell, Ernst Strungmann Inst ESI Neurosci Cooperat, Frankfurt, Germany

[3] Univ Illinois, Dept Psychol, Champaign, IL USA

[4] Univ Sussex, Sch Engn & Informat, Dept Informat, Brighton, E Sussex, England

[5] Aston Univ, Sch Psychol, Birmingham, W Midlands, England

[6] Univ Sydney, Sch Psychol, Sydney, NSW, Australia

[7] Univ Auckland, Dept Psychol, Auckland, New Zealand

[8] Justus Liebig Univ Giessen, Dept Psychol, Giessen, Germany

[9] Univ Marburg & Giessen, Ctr Mind Brain & Behav, Giessen, Germany

[10] Univ Arizona, Dept Linguist, Tucson, AZ USA

[11] Indian Stat Inst, Ctr Soft Comp Res, Kolkata, India

[12] Indian Stat Inst, Machine Intelligence Unit, Kolkata, India

[13] IIIT Allahabad, Applied Sci, Prayagraj, Uttarakhand, India

[14] Tufts Univ, Biomed Engn, Medford, MA USA

[15] Justus Liebig Univ, Dept Psychol, Giessen, Germany

[16] York Univ, Dept Psychol, Toronto, ON, Canada

[17] Univ Oxford, Dept Expt Psychol, Oxford, England

[18] MIT, Quest Intelligence, Dept Brain & Cognit Sci, Cambridge, MA USA

[19] MIT, McGovern Inst Brain Res, Cambridge, MA USA

[20] Stanford Univ, Wu Tsai Neurosci Inst, Stanford, CA USA

[21] Univ Tubingen, Tubingen AI Ctr, Tubingen, Germany

[22] Ecole Polytech Fed Lausanne, Lausanne, Switzerland

[23] Univ Calif San Diego, Dept Cognit Sci, San Diego, CA USA

[24] Univ Rochester, Dept Brain & Cognit Sci, Rochester, NY USA

[25] Ben Gurion Univ Negev, Dept Cognit & Brain Sci, Beer Sheva, Israel

[26] Columbia Univ, Zuckerman Mind Brain Behav Inst, New York, NY USA

[27] NYU, Ctr Neural Sci, New York, NY USA

[28] Univ Glasgow, Sch Psychol & Neurosci, Glasgow, Lanark, Scotland

[29] Max Planck Inst Psycholinguist, Dept Neurobiol Language, Nijmegen, Netherlands

[30] Max Planck Inst Human Cognit & Brain Sci, Leipzig, Germany

[31] Univ Osnabruck, Inst Cognit Sci, Osnabruck, Germany

[32] Columbia Univ, Ctr Sci & Soc, Presidential Scholars Soc & Neurosci, New York, NY USA

[33] Columbia Univ, Italian Acad Adv Studies Amer, New York, NY USA

[34] Harvard Univ, Dept Psychol & Ctr Brain Sci, Cambridge, MA USA

[35] Donders Inst Brain Cognit & Behav, Nijmegen, Netherlands

[36] Univ Penn, Dept Bioengn & Neurosci, Philadelphia, PA USA

[37] CIFAR, Learning Machines & Brains Program, Toronto, ON, Canada

[38] Mila, Montreal, PQ, Canada

[39] McGill Univ, Sch Comp Sci, Montreal, PQ, Canada

[40] McGill Univ, Dept Neurol & Neurosurg, Montreal, PQ, Canada

[41] Montreal Neurol Inst, Montreal, PQ, Canada

[42] NYU, Dept Psychol, New York, NY USA

[43] NYU, Ctr Data Sci, New York, NY USA

[44] Columbia Univ, Dept Psychol Neurosci & Elect Engn, New York, NY USA

[45] Technion, Dept Biomed Engn, Haifa, Israel

[46] Google DeepMind, Mountain View, CA USA

[47] MIT, McGovern Inst, Cambridge, MA USA

[48] Google Res, Mountain View, CA USA

[49] Univ Colorado, Dept Psychol & Neurosci, Boulder, CO USA

[50] Univ Bristol, Dept Comp Sci, Bristol, Avon, England

来源：

BEHAVIORAL AND BRAIN SCIENCES | 2023年 / 46卷

关键词：

D O I：

10.1017/S0140525X23001589

中图分类号：

B84 [心理学];

学科分类号：

04 ; 0402 ;

摘要：

Bowers et al. argue that deep neural networks (DNNs) are poor models of biological vision because they often learn to rival human accuracy by relying on strategies that differ markedly from those of humans. We show that this problem is worsening as DNNs are becoming larger-scale and increasingly more accurate, and prescribe methods for building DNNs that can reliably model biological vision.

引用

页数：77

共 50 条

[1] Learning Not to Learn: Training Deep Neural Networks with Biased Data
Kim, Byungju
Kim, Hyunwoo
Kim, Kyungsu
Kim, Sungjin
Kim, Junmo
[J]. 2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, : 9004 - 9012
[2] Appropriate Learning Rates of Adaptive Learning Rate Optimization Algorithms for Training Deep Neural Networks
Iiduka, Hideaki
[J]. IEEE TRANSACTIONS ON CYBERNETICS, 2022, 52 (12) : 13250 - 13261
[3] A survey on parallel training algorithms for deep neural networks
Yook, Dongsuk
Lee, Hyowon
Yoo, In-Chul
[J]. JOURNAL OF THE ACOUSTICAL SOCIETY OF KOREA, 2020, 39 (06): : 505 - 514
[4] Better schedules for low precision training of deep neural networks
Wolfe, Cameron R.
Kyrillidis, Anastasios
[J]. MACHINE LEARNING, 2024, 113 (06) : 3569 - 3587
[5] Better schedules for low precision training of deep neural networks
Cameron R. Wolfe
Anastasios Kyrillidis
[J]. Machine Learning, 2024, 113 : 3569 - 3587
[6] Towards Better Performance with Heterogeneous Training Data in Acoustic Modeling Using Deep Neural Networks
Huang, Yan
Slaney, Malcolm
Seltzer, Michael L.
Gong, Yifan
[J]. 15TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2014), VOLS 1-4, 2014, : 845 - 849
[7] Training neural networks with harmony search algorithms for classification problems
Kulluk, Sinem
Ozbakir, Lale
Baykasoglu, Adil
[J]. ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE, 2012, 25 (01) : 11 - 19
[8] Deep Learning Neural Networks and Bayesian Neural Networks in Data Analysis
Chernoded, Andrey
Dudko, Lev
Myagkov, Igor
Volkov, Petr
[J]. XXIII INTERNATIONAL WORKSHOP HIGH ENERGY PHYSICS AND QUANTUM FIELD THEORY (QFTHEP 2017), 2017, 158
[9] Training Deep Neural Networks with Constrained Learning Parameters
Date, Prasanna
Carothers, Christopher D.
Mitchell, John E.
Hendler, James A.
Magdon-Ismail, Malik
[J]. 2020 INTERNATIONAL CONFERENCE ON REBOOTING COMPUTING (ICRC 2020), 2020, : 107 - 115
[10] Learning Face Recognition from Limited Training Data using Deep Neural Networks
Peng, Xi
Ratha, Nalini
Pankanti, Sharathchandra
[J]. 2016 23RD INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR), 2016, : 1442 - 1447

← 1 2 3 4 5 →