Exploring loss function topology with cyclical learning rates.

Leslie N. Smith Nicholay Topin

Published in: CoRR (2017)

Keyphrases

loss function
learning rate
reproducing kernel hilbert space
covering numbers
uniform convergence
gaussian kernels
risk minimization
pairwise
learning algorithm
convergence rate
support vector
convergence speed
vc dimension
weight vector
boosting algorithms
least squares
training set