Understanding Approximate Fisher Information for Fast Convergence of Natural Gradient Descent in Wide Neural Networks.

Ryo Karakida Kazuki Osawa

Published in: CoRR (2020)

Keyphrases

fisher information
neural network
probability density function
pattern recognition
information geometry
cost function
convergence rate
positive definite
euclidean space
nearest neighbor
maximum likelihood
noise level
probability distribution
geodesic distance
multidimensional scaling