Prevalence of neural collapse during the terminal phase of deep learning training

Journal: Proceedings of the National Academy of Sciences of the United States of America

Published: 2020-09-21

DOI: 10.1073/pnas.2015509117

Affiliations: 2

Authors: 3

Institutions	Authors	Share
Stanford Department of Statistics, United States of America (USA)		0.67
School of Operations Research and Information Engineering, Cornell University, United States of America (USA)		0.33

Return

Latest supplement

Nature Index 2020 Artificial Intelligence

Artificial intelligence is one of the most rapidly advancing and controversial topics in scientific research. As investments and revenues soar year-on-year and nations vie for leadership in the field, AI research output continues its steep, upward trajectory. But there are significant ethical and technical challenges to overcome.

Access free

Research highlights of partners

Defying the laws of attraction

Keio University

22 Jan 2018

Homing in on an elusive asteroid crater

Curtin University

26 Dec 2018

Sexy selfies fuelled by status anxiety

University of New South Wales (UNSW Sydney)

27 Feb 2020

Research highlights

Sign up for the free weekly Nature Index newsletter
Sign up