PAC-Bayesian Analysis of Martingales and Multiarmed Bandits
Yevgeny SeldinFrançois LavioletteJohn Shawe-TaylorJan PetersPeter AuerPublished in: CoRR (2011)
Keyphrases
- bayesian analysis
- computationally feasible
- stochastic systems
- sample complexity
- upper bound
- pac learning
- probability theory
- multi armed bandits
- personalized ranking
- pac bayesian
- statistically sound
- sample size
- variational bayesian inference
- vc dimension
- noise tolerant
- agnostic learning
- mistake bound
- data sets
- lower bound
- regret bounds
- statistical queries
- graphical models
- dynamic programming
- learning algorithm