arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1107.4506stat.ML
收藏

任意时间Bandit策略的鲁棒性

Robustness of Anytime Bandit Policies

Antoine Salomon, Jean-Yves Audibert

摘要:

研究了随机多臂强盗问题中后悔偏差的存在性。当游戏的总次数n事先被代理人知道时,Audibert et al.(2009)展示了一个策略,使得概率至少为1-1/n,策略的后悔次数为log(n)。他们还表明,Auer et al.(2002)流行的ucb 1政策并不具有这样的性质。这项工作首先回答了一个悬而未决的问题:它扩展了这个负面的结果,任何时候的政策。本文的第二个贡献是设计任何时候都强大的政策,具体的多臂强盗问题,其中的一些限制的可能分布的不同的武器。

英文摘要:

This paper studies the deviations of the regret in a stochastic multi-armed bandit problem. When the total number of plays n is known beforehand by the agent, Audibert et al. (2009) exhibit a policy such that with probability at least 1-1/n, the regret of the policy is of order log(n). They have also shown that such a property is not shared by the popular ucb1 policy of Auer et al. (2002). This work first answers an open question: it extends this negative result to any anytime policy. The second contribution of this paper is to design anytime robust policies for specific multi-armed bandit problems in which some restrictions are put on the set of possible distributions of the different arms.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑