Login / Signup
Critical Point-Finding Methods Reveal Gradient-Flat Regions of Deep Network Losses.
Charles G. Frye
James Simon
Neha S. Wadia
Andrew Ligeralde
Michael Robert DeWeese
Kristofer E. Bouchard
Published in:
Neural Comput. (2021)
Keyphrases
</>
significant improvement
multiscale
critical points
data sets
image processing
bayesian networks
pattern recognition
object recognition
computational cost
peer to peer
network model
gradient method
deep structure