Woodroofe’s one-armed bandit problem revisited
Goldenshluger, Alexander ; Zeevi, Assaf
Ann. Appl. Probab., Tome 19 (2009) no. 1, p. 1603-1633 / Harvested from Project Euclid
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.
Publié le : 2009-08-15
Classification:  Sequential allocation,  online learning,  estimation,  bandit problems,  regret,  inferior sampling rate,  minimax,  rate-optimal policy,  62L05,  60G40,  62C20
@article{1248700629,
     author = {Goldenshluger, Alexander and Zeevi, Assaf},
     title = {Woodroofe's one-armed bandit problem revisited},
     journal = {Ann. Appl. Probab.},
     volume = {19},
     number = {1},
     year = {2009},
     pages = { 1603-1633},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1248700629}
}
Goldenshluger, Alexander; Zeevi, Assaf. Woodroofe’s one-armed bandit problem revisited. Ann. Appl. Probab., Tome 19 (2009) no. 1, pp.  1603-1633. http://gdmltest.u-ga.fr/item/1248700629/