Corpus ID: 7299259

PAC-Bayesian Analysis of Martingales and Multiarmed Bandits

  title={PAC-Bayesian Analysis of Martingales and Multiarmed Bandits},
  author={Yevgeny Seldin and F. Laviolette and J. Shawe-Taylor and Jan Peters and P. Auer},
  • Yevgeny Seldin, F. Laviolette, +2 authors P. Auer
  • Published 2011
  • Computer Science, Mathematics
  • ArXiv
  • We present two alternative ways to apply PAC-Bayesian analysis to sequences of dependent random variables. The first is based on a new lemma that enables to bound expectations of convex functions of certain dependent random variables by expectations of the same functions of independent Bernoulli random variables. This lemma provides an alternative tool to Hoeffding-Azuma inequality to bound concentration of martingale values. Our second approach is based on integration of Hoeffding-Azuma… CONTINUE READING
    PAC-Bayes-Bernstein Inequality for Martingales and its Application to Multiarmed Bandits
    • 12
    • PDF
    (APPRENTISSAGE SÉQUENTIEL : Bandits, Statistique et Renforcement
    • 6
    • PDF
    PAC-Bayesian Analysis of the Exploration-Exploitation Trade-off
    • 7
    • PDF
    Regulating Greed Over Time
    • 4
    • PDF


    Publications referenced by this paper.
    Finite-time Analysis of the Multiarmed Bandit Problem
    • 4,137
    • PDF
    The Nonstochastic Multiarmed Bandit Problem
    • 1,561
    • PDF
    PAC-Bayesian Generalisation Error Bounds for Gaussian Process Classification
    • 184
    • PDF
    Reinforcement Learning: An Introduction
    • 25,507
    • PDF
    Bayesian Gaussian process models : PAC-Bayesian generalisation error bounds and sparse approximations
    • 170
    • PDF
    A PAC analysis of a Bayesian estimator
    • 105
    Elements of Information Theory
    • 38,856
    • Highly Influential
    • PDF
    Simplified PAC-Bayesian Margin Bounds
    • 148
    • PDF
    Chromatic PAC-Bayes Bounds for Non-IID Data
    • 62
    • PDF
    UCB revisited: Improved regret bounds for the stochastic multi-armed bandit problem
    • 177
    • PDF