Tighter PAC-Bayes bounds through distribution-dependent priors.

Guy Lever François Laviolette John Shawe-Taylor

Published in: Theor. Comput. Sci. (2013)

Keyphrases

pac bayes
risk bounds
upper bound
generalization bounds
lower bound
data dependent
linear classifiers
heavy tailed
large deviations
empirical risk minimization
worst case
decision theoretic
vc dimension
learning algorithm
bayesian framework
special case
pairwise