Gradient descent almost surely avoids strict saddles for step sizes alpha < 2/L, provided the Hessian eigenvalue alpha^-1 occurs only on a measure-zero set.
Escaping fr om saddle points–online stochas- tic gradient for tensor decomposition
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Extending the step-size restriction for gradient descent to avoid strict saddle points
Gradient descent almost surely avoids strict saddles for step sizes alpha < 2/L, provided the Hessian eigenvalue alpha^-1 occurs only on a measure-zero set.