Woodroofe's one-armed bandit problem revisited
Authors:
- Alexander GoldenshlugerAGAlexander Goldenshluger
- Assaf ZeeviAZAssaf Zeevi
Preprint
Abstract
We consider the one-armed bandit problem of Woodroofe [J. Amer. Statist. Assoc. 74 (1979) 799--806], which involves sequential sampling from two populations: one whose characteristics are known, and one which depends on an unknown parameter and incorporates a covariate. The goal is to maximize cumulative expected reward. We study this problem in a minimax setting, and develop rate-optimal polices that involve suitable modifications of the myopic rule. It is shown that the regret, as well as the rate of sampling from the inferior population, can be finite or grow at various rates with the time horizon of the problem, depending on "local" properties of the covariate distribution. Proofs rely on martingale methods and information theoretic arguments.Keywords
All Related Versions
- Version 1, 2009-09-01, ArXiv
- Published version: The Annals of Applied Probability, 19 (4), 1603.
This publication has 10 references indexed in Scilit:
- Pseudo-maximization and self-normalized processesProbability Surveys, 2007
- Bandit problems with side observationsIEEE Transactions on Automatic Control, 2005
- Self-normalized processes: exponential inequalities, moment bounds and iterated logarithm lawsThe Annals of Probability, 2004
- Optimal aggregation of classifiers in statistical learningThe Annals of Statistics, 2004
- Randomized Allocation with nonparametric estimation for a multi-armed bandit problem with covariatesThe Annals of Statistics, 2002
- Deviation probability bound for martingales with applications to statistical estimationStatistics & Probability Letters, 2000
- Smooth discrimination analysisThe Annals of Statistics, 1999
- One-Armed Bandit Problems with CovariatesThe Annals of Statistics, 1991
- Information Inequalities for the Bayes RiskThe Annals of Statistics, 1990
- Some Aspects of the Sequential Design of Experiments1985