Fixing the problems of deep neural networks will require better training data and learning algorithms
Bowers, Jeffrey S.; Malhotra, Gaurav; Dujmovic, Marin; Montero, Milton Llera; Tsvetkov, Christian; Biscione, Valerio; Puebla, Guillermo; Adolfi, Federico; Hummel, John E.; Heaton, Rachel F.; Evans, Benjamin D.; Mitchell, Jeffrey; Blything, Ryan; Anderson, Barton L.; Storrs, Katherine R.; Fleming, Roland W.; Bever, Thomas G.; Chomsky, Noam; Fong, Sandiway; Piattelli-Palmarini, Massimo; Chandran, Keerthi S.; Paul, Amrita Mukherjee; Paul, Avijit; Ghosh, Kuntal; de Vries, Jelmer Philip; Flachot, Alban; Morimoto, Takuma; Gegenfurtner, Karl R.; DiCarlo, James J.; Yamins, Daniel L. K.; Ferguson, Michael E.; Fedorenko, Evelina; Bethge, Matthias; Bonnen, Tyler; Schrimpf, Martin; German, Joseph Scott; Jacobs, Robert A.; Golan, Tal; Taylor, JohnMark; Schutt, Heiko; Peters, Benjamin; Sommers, Rowan P.; Seeliger, Katja; Doerig, Adrien; Linton, Paul; Konkle, Talia; van Gerven, Marcel; Kording, Konrad; Richards, Blake; Kietzmann, Tim C.; Lindsay, Grace W.; Kriegeskorte, Nikolaus; Gur, Moshe; Hermann, Katherine; Nayebi, Aran; van Steenkiste, Sjoerd; Jones, Matt; Houghton, Conor; Kazanina, Nina; Sukumaran, Priyanka; Kellman, Philip J.; Baker, Nicholas; Garrigan, Patrick; Phillips, Austin; Lu, Hongjing; Koculak, Marcin; Wierzchon, Michal; Li, Aedan Y.; Mur, Marieke; Lin, Hause; Linsley, Drew; Serre, Thomas; Liu, Jianghao; Bartolomeo, Paolo; Love, Bradley C.; Mok, Robert M.; Moldoveanu, Mihnea; Op de Beeck, Hans; Bracci, Stefania; Rothkopf, Constantin; Bremmer, Frank; Fiehler, Katja; Dobs, Katharina; Triesch, Jochen; Slagter, Heleen A.; Spratling, Michael W.; Srivastava, Nisheeth; Sifar, Anjali; Srinivasan, Narayanan; Summerfield, Christopher; Thompson, Jessica A. F.; Tarr, Michael J.; Veit, Walter; Browning, Heather; Wichmann, Felix A.; Kornblith, Simon; Geirhos, Robert; Xu, Yaoda; Vaziri-Pashkam, Maryam; Yovel, Galit; Abudarham, Naphtali
2023

Abstract

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.

Details

Title Fixing the problems of deep neural networks will require better training data and learning algorithms
Author(s) Bowers, Jeffrey S. ; Malhotra, Gaurav ; Dujmovic, Marin ; Montero, Milton Llera ; Tsvetkov, Christian ; Biscione, Valerio ; Puebla, Guillermo ; Adolfi, Federico ; Hummel, John E. ; Heaton, Rachel F. ; Evans, Benjamin D. ; Mitchell, Jeffrey ; Blything, Ryan ; Anderson, Barton L. ; Storrs, Katherine R. ; Fleming, Roland W. ; Bever, Thomas G. ; Chomsky, Noam ; Fong, Sandiway ; Piattelli-Palmarini, Massimo ; Chandran, Keerthi S. ; Paul, Amrita Mukherjee ; Paul, Avijit ; Ghosh, Kuntal ; de Vries, Jelmer Philip ; Flachot, Alban ; Morimoto, Takuma ; Gegenfurtner, Karl R. ; DiCarlo, James J. ; Yamins, Daniel L. K. ; Ferguson, Michael E. ; Fedorenko, Evelina ; Bethge, Matthias ; Bonnen, Tyler ; Schrimpf, Martin ; German, Joseph Scott ; Jacobs, Robert A. ; Golan, Tal ; Taylor, JohnMark ; Schutt, Heiko ; Peters, Benjamin ; Sommers, Rowan P. ; Seeliger, Katja ; Doerig, Adrien ; Linton, Paul ; Konkle, Talia ; van Gerven, Marcel ; Kording, Konrad ; Richards, Blake ; Kietzmann, Tim C. ; Lindsay, Grace W. ; Kriegeskorte, Nikolaus ; Gur, Moshe ; Hermann, Katherine ; Nayebi, Aran ; van Steenkiste, Sjoerd ; Jones, Matt ; Houghton, Conor ; Kazanina, Nina ; Sukumaran, Priyanka ; Kellman, Philip J. ; Baker, Nicholas ; Garrigan, Patrick ; Phillips, Austin ; Lu, Hongjing ; Koculak, Marcin ; Wierzchon, Michal ; Li, Aedan Y. ; Mur, Marieke ; Lin, Hause ; Linsley, Drew ; Serre, Thomas ; Liu, Jianghao ; Bartolomeo, Paolo ; Love, Bradley C. ; Mok, Robert M. ; Moldoveanu, Mihnea ; Op de Beeck, Hans ; Bracci, Stefania ; Rothkopf, Constantin ; Bremmer, Frank ; Fiehler, Katja ; Dobs, Katharina ; Triesch, Jochen ; Slagter, Heleen A. ; Spratling, Michael W. ; Srivastava, Nisheeth ; Sifar, Anjali ; Srinivasan, Narayanan ; Summerfield, Christopher ; Thompson, Jessica A. F. ; Tarr, Michael J. ; Veit, Walter ; Browning, Heather ; Wichmann, Felix A. ; Kornblith, Simon ; Geirhos, Robert ; Xu, Yaoda ; Vaziri-Pashkam, Maryam ; Yovel, Galit ; Abudarham, Naphtali
Published in Behavioral And Brain Sciences
Volume 46
Pages e400
Date 2023-12-06
Publisher Cambridge Univ Press, Cambridge
ISSN 0140-525X
1469-1825
DOI https://doi.org/10.1017/S0140525X23001589
Other identifier(s) View record in Web of Science
Laboratories UPSCHRIMPF1
Record Appears in Scientific production and competences > SV - School of Life Sciences > INX-SV - Interschool Institute Neuro-X (SV) > UPSCHRIMPF1 - Prof. Schrimpf Group - SV
Peer-reviewed publications
Work produced at EPFL
Journal Articles
Published
Grant ONR: N00014-19-1-2029
NSF: IIS-1912280
ANR-3IA Artificial and Natural Intelligence Toulouse Institute: ANR-19-PI3A-0004
Record creation date 2024-02-21