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 1102.2490math.STcs.LGcs.SYeess.SYmath.OCstat.TH
收藏

KL-UCB算法:有界随机赌博机及其扩展

The KL-UCB Algorithm for Bounded Stochastic Bandits and Beyond

LTCI(LTCI实验室);CNRS(法国国家科学研究中心);Telecom ParisTech(巴黎高等电信学院)

Aurélien Garivier, Olivier Cappé

AI总结:

本文提出KL-UCB算法,通过有限时间分析证明其在有界奖励下优于UCB和UCB2,在伯努利奖励下达到Lai和Robbins下界,并扩展到指数族分布,数值实验显示其高效稳定。

摘要:

本文对KL-UCB算法进行了有限时间分析,该算法是一种在线、无时间视界的随机赌博机问题的索引策略。我们证明了两个不同的结果:首先,对于任意有界奖励,KL-UCB算法满足比UCB或UCB2一致更优的遗憾界;其次,在伯努利奖励的特殊情况下,它达到了Lai和Robbins的下界。此外,我们展示了KL-UCB算法的简单改编对于特定类别的(可能无界)奖励也是最优的,包括那些从指数族分布生成的奖励。一项大规模数值研究将KL-UCB与其主要竞争对手(UCB、UCB2、UCB-Tuned、UCB-V、DMED)进行比较,表明KL-UCB非常高效且稳定,包括在短时间范围内。KL-UCB也是唯一始终优于基本UCB策略的方法。我们的遗憾界依赖于附录中陈述并证明的具有独立兴趣的偏差结果。作为副产品,我们还获得了标准UCB算法的改进遗憾界。

英文摘要:

This paper presents a finite-time analysis of the KL-UCB algorithm, an online, horizon-free index policy for stochastic bandit problems. We prove two distinct results: first, for arbitrary bounded rewards, the KL-UCB algorithm satisfies a uniformly better regret bound than UCB or UCB2; second, in the special case of Bernoulli rewards, it reaches the lower bound of Lai and Robbins. Furthermore, we show that simple adaptations of the KL-UCB algorithm are also optimal for specific classes of (possibly unbounded) rewards, including those generated from exponential families of distributions. A large-scale numerical study comparing KL-UCB with its main competitors (UCB, UCB2, UCB-Tuned, UCB-V, DMED) shows that KL-UCB is remarkably efficient and stable, including for short time horizons. KL-UCB is also the only method that always performs better than the basic UCB policy. Our regret bounds rely on deviations results of independent interest which are stated and proved in the Appendix. As a by-product, we also obtain an improved regret bound for the standard UCB algorithm.

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。

↑