Search NASA⌕ Search
Home/Archive/DOE OSTI

DOE OSTI · 1982997

slimTrain---A Stochastic Approximation Method for Training Separable Deep Neural Networks

Newman, Elizabeth (ORCID:0000000263097706)·Chung, Julianne (ORCID:0000000267604736)·Chung, Matthias (ORCID:0000000178224539)·Ruthotto, Lars (ORCID:0000000308033299)
View original source ↗Export metadata ↓

Abstract

Not provided.

Explore related subjects

Mathematics

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Newman, Elizabeth (ORCID:0000000263097706), Chung, Julianne (ORCID:0000000267604736), Chung, Matthias (ORCID:0000000178224539), Ruthotto, Lars (ORCID:0000000308033299). 2022-08-01. slimTrain---A Stochastic Approximation Method for Training Separable Deep Neural Networks. https://doi.org/10.1137/21m1452512

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related reports

DOE OSTI2024

Global Magni4icence, or: 4G Networks

The global magnificent four theory is the homological version of a maximally supersymmetric $(8+1)$-dimensional gauge theory on a Calabi-Yau fourfold fibered over a circle. In the case of a toric fourfold we conjecture the formula for its twisted Witten index. String-theoretically we count the BPS states of a system of $D0$-$D2$-$D4$-$D6$-$D8$-branes on the Calabi-Yau fourfold in the presence of a large Neveu-Schwarz $B$-field. Mathematically, we develop the equivariant $K$-theoretic DT4 theory, by constructing the four-valent vertex with generic plane partition asymptotics. Physically, the vertex is a supersymmetric localization of a non-commutative gauge theory in $8+1$ dimensions.

Mathematics↗