Asymptotic minimaxity of false discovery rate thresholding for sparse exponential data

Asymptotic minimaxity of false discovery rate thresholding for sparse exponential data
复制标题

DOI:
10.1214/009053606000000920
复制
发表时间:
2006-12-01
影响因子:
4.5
通讯作者:
Jin, Jiashun
Jin, Jiashun
中科院分区:
数学1区
文献类型:
--
作者:
Donoho, David;Jin, Jiashun

文献摘要

被引文献

相似文献

我们将FDR阈值化应用于坐标为Xi,i = 1,.,n与单个平均值mu(i)是独立指数。向量mu =(mu(i))被认为是稀疏的,大多数坐标为1,但有一小部分明显大于1;粗略地说,大多数坐标只是“噪音”,但有一小部分包含“信号”。我们在log(mu(i))的恢复过程中用全坐标均方误差来度量风险,并研究了log(mu(i)),1/n Sigma(n)(i=1)log(p)(mu(i))的全坐标p-范数约束下的参数空间上的Minimax估计
We apply FDR thresholding to a non-Gaussian vector whose coordinates X-i, i = 1,..., n, are independent exponential with individual means mu(i). The vector mu = (mu(i)) is thought to be sparse, with most coordinates 1 but a small fraction significantly larger than 1; roughly, most coordinates are simply 'noise,' but a small fraction contain 'signal.' We measure risk by percoordinate mean-squared error in recovering log(mu(i)), and study minimax estimation over parameter spaces defined by constraints on the per-coordinate p-norm of log(mu(i)), 1/n Sigma(n)(i=1) log(p) (mu(i))