Optimistic Gittins Indices

We propose a tightening sequence of optimistic approximations to the Gittins index in “Optimistic Indices.” show that use these concert with an increasing discount factor appears offer compelling alternative state-of-the-art schemes proposed for Bayesian multiarmed bandit problem. prove indices constitutes regret optimal algorithm. Perhaps more interestingly, even loosest substantial performance improvements over alternatives while incurring little no additional computational overhead relative simplest alternatives.