Finite-time analysis of the multiarmed bandit problem-英文文献.pdf

Machine Learning, 47, 235–256, 2002 c 2002 Kluwer Academic Publishers. Manufactured in The Netherlands. Finite-time Analysis of the Multiarmed Bandit Problem* PETER AUER pauer@igi.tu-graz.ac.at University of Technology Graz, A-8010 Graz, Austria ` NICOLO CESA-BIANCHI cesa-bianchi@dti.unimi.it DTI, University of Milan, via Bramante 65, I-26013 Crema, Italy PAUL FISCHER fischer@ls2.informatik.uni-dortmund.de ¨ Lehrstuhl Informatik II, Universitat Dortmund, D-44221 Dortmund, Germany Editor: Jyrki Kivinen Abstract. Reinforcement learning policies face the exploration versus exploitation dilemma, i.e. the search for a balance between exploring the environment to find profitable actions while taking the empirically best action as often as possible. A popular measure of a policy’s success in addressing this dilemma is the regret, that is the loss due to the fact that the globally optimal policy is not followed all the times. One of the simplest examples of the exploration/exploitation dilemma is the multi-armed bandit problem. Lai and Robbins were the first ones to show that the regret for this problem has to grow at least logarithmically in the number of plays. Since then, policies which asymptotically achieve this regret have been devised by Lai and Robbins and many others. In this work we show that the optimal logarithmic regret is also achievable uniformly over time, with simple and efficient policies, and for all reward distributions with bounded support. Keywords: bandit problems, adaptive allocation rules, finite horizo