Performance Guarantees for Policy Learning.

Performance Guarantees for Policy Learning.
复制标题

DOI:
10.1214/19-aihp1034
复制
发表时间:
2020-08
期刊:
Annales de l'I.H.P. Probabilites et statistiques
影响因子:
--
通讯作者:
Chambaz A
Chambaz A
中科院分区:
其他
文献类型:
--
作者:
Luedtke A;Chambaz A

文献摘要

参考文献

被引文献

相似文献

本文给出了最优策略估计中后悔衰减的性能保证。我们给出了一个保证金免费的结果表明,估计类内最优政策的遗憾衰减是二阶经验风险最小化的Donsker类时,数据是从一个固定的数据分布,不随样本大小而变化,遗憾衰减的速度比标准误差的有效估计值的最优政策。我们还提出了一个结果,保证政策估计的遗憾衰减的情况下,政策福尔斯属于一个受限制的类和数据产生的局部扰动的一个固定的分布,这种保证是均匀的局部扰动的方向。最后,我们给出了一个结果,从分类文献表明,更快的后悔衰减是可能的,通过插件估计提供了保证金条件成立。考虑三个例子。在这些例子中,后悔是用平均值或中值来表示的,可能的行动数量是两个或两个以上。
This article gives performance guarantees for the regret decay in optimal policy estimation. We give a margin-free result showing that the regret decay for estimating a within-class optimal policy is second-order for empirical risk minimizers over Donsker classes when the data are generated from a fixed data distribution that does not change with sample size, with regret decaying at a faster rate than the standard error of an efficient estimator of the value of an optimal policy. We also present a result giving guarantees on the regret decay of policy estimators for the case that the policy falls within a restricted class and the data are generated from local perturbations of a fixed distribution, where this guarantee is uniform in the direction of the local perturbation. Finally, we give a result from the classification literature that shows that faster regret decay is possible via plug-in estimation provided a margin condition holds. Three examples are considered. In these examples, the regret is expressed in terms of either the mean value or the median value, and the number of possible actions is either two or finitely many.
DOI: 10.1214/10-aos864
发表时间: 2011-04-01
影响因子: 4.5
作者:
Qian M;Murphy SA
通讯作者: Murphy SA
DOI: 10.3982/ecta6630
发表时间: 2009-09-01
期刊: ECONOMETRICA
影响因子: 6.1
作者:
Hirano, Keisuke;Porter, Jack R.
通讯作者: Porter, Jack R.
DOI: 10.1214/16-aos1534
发表时间: 2017-12-01
影响因子: 4.5
作者:
Chambaz, Antoine;Zheng, Wenjing;van der Laan, Mark J.
通讯作者: van der Laan, Mark J.
DOI: 10.1214/009053606000001217
发表时间: 2007-04-01
影响因子: 4.5
作者:
Audibert, Jean-Yves;Tsybakov, Alexandre B.
通讯作者: Tsybakov, Alexandre B.
DOI: 10.1515/1557-4679.1423
发表时间: 2012-07-20
期刊: The international journal of biostatistics
影响因子: --
作者:
Rubin, Daniel B;van der Laan, Mark J
通讯作者: van der Laan, Mark J