Lower error bounds for the stochastic gradient descent optimization algorithm: Sharp convergence rates for slowly and fast decaying learning rates

Thumbnail
Metadata only
Author
Jentzen, Arnulf
von Wurstemberger, Philippe
Date
2020-04
Type
    Journal Article
Publication status
published
External links
Journal / series
Journal of Complexity
Volume
57
Pages / Article No.
101438
Publisher
Elsevier
Subject
SGD; Lower bounds; Machine learning; Error analysis
Organisational unit
03951 - Jentzen, Arnulf (ehemalig) / Jentzen, Arnulf (former)  
02501 - Seminar für Angewandte Mathematik / Seminar for Applied Mathematics  
02204 - RiskLab / RiskLab  
Funding
175699 - Higher order numerical approximation methods for stochastic partial differential equations (SNF)
Related publications and datasets
More
Show all metadata
Citations
Cited 6 times in Web of Science
Cited 6 times in Scopus
ETH Bibliography
yes
Altmetrics