Technical perspective: Why don't today's deep nets overfit to their training data?

Author(s): Arora, Sanjeev

Download

To refer to this page use: http://arks.princeton.edu/ark:/88435/pr10c2q

Full metadata record

DC Field	Value	Language
dc.contributor.author	Arora, Sanjeev	-
dc.date.accessioned	2021-10-08T19:50:39Z	-
dc.date.available	2021-10-08T19:50:39Z	-
dc.date.issued	2021	en_US
dc.identifier.citation	Arora, Sanjeev. "Technical perspective: Why don't today's deep nets overfit to their training data?." Communications of the ACM 64, no. 3 (2021): pp. 106. doi:10.1145/3446773	en_US
dc.identifier.issn	0001-0782	-
dc.identifier.uri	http://arks.princeton.edu/ark:/88435/pr10c2q	-
dc.description.abstract	THE FOLLOWING ARTICLE by Zhang et al. is well-known for having highlighted that widespread success of deep learn- ing in artificial intelligence brings with it a fundamental new theoretical challenge, specifically: Why don’t to- day’s deep nets overfit to training data? This question has come to animate the theory of deep learning.	en_US
dc.format.extent	106	en_US
dc.language.iso	en_US	en_US
dc.relation.ispartof	Communications of the ACM	en_US
dc.rights	Final published version. Article is made available in OAR by the publisher's permission or policy.	en_US
dc.title	Technical perspective: Why don't today's deep nets overfit to their training data?	en_US
dc.type	Journal Article	en_US
dc.identifier.doi	10.1145/3446773	-
pu.type.symplectic	http://www.symplectic.co.uk/publications/atom-terms/1.0/journal-article	en_US

Files in This Item:

File	Description	Size	Format
DeepNetNoOverfit.pdf		437.12 kB	Adobe PDF	View/Download