Abstract
The stochastic multi-armed bandit model is a simple abstraction that has proven useful in many different contexts in statistics and machine learning. Whereas the achievable limit in terms of regret minimization is now well known, our aim is to contribute to a better understanding of the performance in terms of identifying the m-best arms. We introduce generic notions of complexity for the two dominant frameworks considered in the literature: fixed-budget and fixed-condence settings. In the fixed-confidence setting, we provide the first known distribution-dependent lower bound on the complexity that involves information-theoretic quantities and holds when m>=1 under general assumptions. In the specific case of two armed-bandits, we derive rened lower bounds in both the fixed-confidence and fixed-budget settings, along with matching algorithms for Gaussian and Bernoulli bandit models. These results show in particular that the complexity of the fixed-budget setting may be smaller than the complexity of the fixed-condence setting, contradicting the familiar behavior observed when testing fully specied alternatives. In addition, we also provide improved sequential stopping rules that have guaranteed error probabilities and shorter average running times. The proofs rely on two technical results that are of independent interest : a deviation lemma for self-normalized sums, and a novel change of measure inequality for bandit models.
Joint work with Emilie Kaufmann and Olivier Cappé.