arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1601.06259stat.MLcs.ITcs.LGmath.ITmath.STstat.TH
收藏

线性独立性检验的Minimax下界

Minimax Lower Bounds for Linear Independence Testing

University of California, Berkeley(加州大学伯克利分校);Carnegie Mellon University(卡内基梅隆大学)

Aaditya Ramdas, David Isenberg, Aarti Singh, Larry Wasserman

摘要:

线性独立性检验是一个基本的信息理论和统计问题,可以提出如下:给定$n$点$\{(X_i,Y_i)\}^n_{i=1}$从$p+q$多维分布,其中$X_i \in \mathbb{R}^p$和$Y_i \in\mathbb{R}^q$,确定是否$a^T X$和$b^T Y$是不相关的每个$a \in \mathbb{R}^p, b\in \mathbb{R}^q$或没有。我们给这个问题的极大极小下界(当$p+q,n \to \infty$,$(p+q)/n \leq κ< \infty$,没有稀疏性假设)。总之,我们的结果意味着,对于任何程序(测试),$n$必须至少与$\sqrt {pq}/\|Σ_{XY}\|_F^2$一样大才具有非平凡功效,其中$Σ_{XY}$是$X,Y$的互协方差矩阵。我们还提供了一些证据,下界是紧的,连接到两个样本测试和回归在特定的设置。

英文摘要:

Linear independence testing is a fundamental information-theoretic and statistical problem that can be posed as follows: given $n$ points $\{(X_i,Y_i)\}^n_{i=1}$ from a $p+q$ dimensional multivariate distribution where $X_i \in \mathbb{R}^p$ and $Y_i \in\mathbb{R}^q$, determine whether $a^T X$ and $b^T Y$ are uncorrelated for every $a \in \mathbb{R}^p, b\in \mathbb{R}^q$ or not. We give minimax lower bound for this problem (when $p+q,n \to \infty$, $(p+q)/n \leq κ< \infty$, without sparsity assumptions). In summary, our results imply that $n$ must be at least as large as $\sqrt {pq}/\|Σ_{XY}\|_F^2$ for any procedure (test) to have non-trivial power, where $Σ_{XY}$ is the cross-covariance matrix of $X,Y$. We also provide some evidence that the lower bound is tight, by connections to two-sample testing and regression in specific settings.

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。

↑