English

Simpler PAC-Bayesian Bounds for Hostile Data

Machine Learning 2019-05-24 v2 Statistics Theory Statistics Theory

Abstract

PAC-Bayesian learning bounds are of the utmost interest to the learning community. Their role is to connect the generalization ability of an aggregation distribution ρ\rho to its empirical risk and to its Kullback-Leibler divergence with respect to some prior distribution π\pi. Unfortunately, most of the available bounds typically rely on heavy assumptions such as boundedness and independence of the observations. This paper aims at relaxing these constraints and provides PAC-Bayesian learning bounds that hold for dependent, heavy-tailed observations (hereafter referred to as \emph{hostile data}). In these bounds the Kullack-Leibler divergence is replaced with a general version of Csisz\'ar's ff-divergence. We prove a general PAC-Bayesian bound, and show how to use it in various hostile settings.

Cite

@article{arxiv.1610.07193,
  title  = {Simpler PAC-Bayesian Bounds for Hostile Data},
  author = {Pierre Alquier and Benjamin Guedj},
  journal= {arXiv preprint arXiv:1610.07193},
  year   = {2019}
}

Comments

18 pages