Sequential Analysis ( IF 0.6 ) Pub Date : 2021-03-11 , DOI: 10.1080/07474946.2021.1847965 Aurélien Garivier 1 , Emilie Kaufmann 2
Abstract
In this article, we study sequential testing problems with overlapping hypotheses. We first focus on the simple problem of assessing if the mean μ of a Gaussian distribution is smaller or larger than a fixed if both answers are considered to be correct. Then, we consider probably approximately correct best arm identification in a bandit model: given K probability distributions on with means we derive the asymptotic complexity of identifying, with risk at most δ, an index such that We provide nonasymptotic bounds on the error of a parallel general likelihood ratio test, which can also be used for more general testing problems. We further propose a lower bound on the number of observations needed to identify a correct hypothesis. Those lower bounds rely on information-theoretic arguments, and specifically on two versions of a change of measure lemma (a high-level form and a low-level form) whose relative merits are discussed.
中文翻译:
重叠假设的非渐近顺序检验应用于匪徒模型中的近乎最佳手臂识别
摘要
在本文中,我们研究具有重叠假设的顺序测试问题。我们首先关注评估高斯分布的均值μ小于或大于固定点的简单问题 如果 这两个答案都被认为是正确的。然后,我们考虑在强盗模型中可能近似正确的最佳手臂识别:给定K个概率分布 用手段 我们得出了风险最大为δ的指标的识别的渐近复杂度 这样 我们提供了关于并行一般似然比检验的误差的非渐近边界,它也可以用于更一般的检验问题。我们进一步提出了确定正确假设所需的观测值数量的下限。这些下限依赖于信息理论论证,特别是依赖于衡量度量引理变化的两个版本(高级形式和低级形式),它们的相对优点在此进行了讨论。