期刊
ANNALS OF STATISTICS
卷 49, 期 4, 页码 2151-2177出版社
INST MATHEMATICAL STATISTICS-IMS
DOI: 10.1214/20-AOS2030
关键词
Conditional independence; minimax optimality; hypothesis testing
资金
- NSF [CCF1763734, DMS1713003]
The study focuses on the problem of conditional independence testing of X and Y given Z, considering smoothness assumptions on conditional distributions and testing difficulty. Lower and upper bounds were derived on the critical radius of separation between null and alternative hypotheses in the total variation metric.
We consider the problem of conditional independence testing of X and Y given Z where X, Y and Z are three real random variables and Z is continuous. We focus on two main cases-when X and Y are both discrete, and when X and Y are both continuous. In view of recent results on conditional independence testing [Ann. Statist. 48 (2020) 1514-1538], one cannot hope to design nontrivial tests, which control the type I error for all absolutely continuous conditionally independent distributions, while still ensuring power against interesting alternatives. Consequently, we identify various, natural smoothness assumptions on the conditional distributions of X, Y vertical bar Z = z as z varies in the support of Z, and study the hardness of conditional independence testing under these smoothness assumptions. We derivematching lower and upper bounds on the critical radius of separation between the null and alternative hypotheses in the total variation metric. The tests we consider are easily implementable and rely on binning the support of the continuous variable Z. To complement these results, we provide a new proof of the hardness result of Shah and Peters.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据