Linking task structure and neural network dynamics

The solutions found by neural networks to solve a task are often inscrutable. We have little insight into why a particular structure emerges in a network. By reverse engineering neural networks from dynamical principles, Dubreuil, Valente et al. show how neural population structure enables computati...

Full description

Saved in:
Bibliographic Details
Published inNature neuroscience Vol. 25; no. 6; pp. 679 - 681
Main Authors Márton, Christian David, Zhou, Siyan, Rajan, Kanaka
Format Journal Article
LanguageEnglish
Published New York Nature Publishing Group US 01.06.2022
Nature Publishing Group
Subjects
631/378/116/1925
631/378/116/2393
Animal Genetics and Genomics
Behavioral Sciences
Biological Techniques
Biomarkers
Biomedical and Life Sciences
Biomedicine
Computer applications
Decision making
Memory, Short-Term
Neural networks
Neural Networks, Computer
Neurobiology
Neurons
Neurosciences
News & Views
news-and-views
Population
Population structure
Reverse engineering
Support vector machines
Online AccessGet full text

Cover

More Information
Summary:The solutions found by neural networks to solve a task are often inscrutable. We have little insight into why a particular structure emerges in a network. By reverse engineering neural networks from dynamical principles, Dubreuil, Valente et al. show how neural population structure enables computational flexibility.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
ObjectType-Commentary-3
content type line 23
Co-authors: cdmarton@gmail.com, szhou@g.harvard.edu
ISSN:1097-6256
1546-1726
DOI:10.1038/s41593-022-01090-w