Fixing the problems of deep neural networks will require better training data and learning algorithms
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.
Cite this item
Bowers, J. S., Malhotra, G., Dujmovic, M., Montero, M. L., Tsvetkov, C., Biscione, V., Puebla, G., Adolfi, F., Hummel, J. E., Heaton, R. F., Evans, B. D., Mitchell, J., Blything, R., Anderson, B. L., Storrs, K. R., Fleming, R. W., Bever, T. G., Chomsky, N., Fong, S., et al. (2023). Fixing the problems of deep neural networks will require better training data and learning algorithms. Behavioral and Brain Sciences, 46, Article e400. https://doi.org/10.1017/S0140525X23001589
Click Copy to copy the citation, or select the text manually.
WOS:001127780700001
2023-12-06
46
e400
REVIEWED
EPFL
| Funder | Grant Number |
ONR | N00014-19-1-2029 |
NSF | IIS-1912280 |
ANR-3IA Artificial and Natural Intelligence Toulouse Institute | ANR-19-PI3A-0004 |