On the existence of global minima and convergence analyses for gradient descent methods in the training of deep neural networksOpen Access

Jentzen A, Riekert A

Research article (journal) | Peer reviewed

Details about the publication

JournalJournal of Machine Learning (J. Mach. Learn.)
Volume1
Issue2
Page range141-246
StatusPublished
Release year2022
Language in which the publication is writtenEnglish
KeywordsDeep learning; Global minimum; Kurdyka-Łojasiewicz inequality; Gradient descent; Optimization

Authors from the University of Münster

Jentzen, Arnulf
Riekert, Adrian