我们提出了部分相关筛选(PCS)作为一个新的逐行的方法来估计一个大的精度矩阵$Ω$。为了估计$Ω$的第$i$行,$1 \leq i \leq p$,PCS使用“筛选”步骤和“清除”步骤。在筛选步骤中,PCS使用逐阶段算法招募索引的(小)子集,其中在每个阶段中,算法通过添加与$i$具有最大(在幅度上)经验偏相关性的索引$j$来更新所招募索引的集合。在清理步骤中,PCS重新调查所有被招募的索引并使用它们来重构$Ω$的第$i$行。PCS在计算上是有效的并且在存储器中是适度的用途:为了估计$Ω$的行,它仅需要经验协方差阵的几行(顺序确定)。这使得PCS能够执行大精度矩阵(例如,$p=10K$),并为估计更大的精度矩阵打开了大门。我们使用PCS进行分类。更高的批评保持(HCT)是一个最近的分类器,享有最优性,但在实践中充分利用其潜力,需要一个很好的估计精度矩阵$Ω$。将HCT与估计$Ω$的任何方法相结合给出了新的分类器:示例包括HCT-PCS和HCT-glasso。我们已经将HCT-PCS应用于两个大的微阵列数据集($p = 8K$和$10K$)进行分类,其中它不仅显著优于HCT-glasso,而且与支持向量机(SVM)和随机森林(RF)竞争。结果表明,PCS给出了更有用的估计比glasso的$Ω$。我们建立了一个一般的理论框架,并表明,在一个广泛的背景下,PCS完全恢复的支持$Ω$和HCT-PCS产生最佳的分类行为。我们的证明揭示了阶段明智的程序的行为有趣的光。
英文摘要:
We propose Partial Correlation Screening (PCS) as a new row-by-row approach to estimating a large precision matrix $Ω$. To estimate the $i$-th row of $Ω$, $1 \leq i \leq p$, PCS uses a Screen step and a Clean step. In the Screen step, PCS recruits a (small) subset of indices using a stage-wise algorithm, where in each stage, the algorithm updates the set of recruited indices by adding the index $j$ that has the largest (in magnitude) empirical partial correlation with $i$. In the Clean step, PCS re-investigates all recruited indices and use them to reconstruct the $i$-th row of $Ω$.
PCS is computationally efficient and modest in memory use: to estimate a row of $Ω$, it only needs a few rows (determined sequentially) of the empirical covariance matrix. This enables PCS to execute the estimation of a large precision matrix (e.g., $p=10K$) in a few minutes, and open doors to estimating much larger precision matrices.
We use PCS for classification. Higher Criticism Thresholding (HCT) is a recent classifier that enjoys optimality, but to exploit its full potential in practice, one needs a good estimate of the precision matrix $Ω$. Combining HCT with any approach to estimating $Ω$ gives a new classifier: examples include HCT-PCS and HCT-glasso. We have applied HCT-PCS to two large microarray data sets ($p = 8K$ and $10K$) for classification, where it not only significantly outperforms HCT-glasso, but also is competitive to the Support Vector Machine (SVM) and Random Forest (RF). The results suggest that PCS gives more useful estimates of $Ω$ than the glasso.
We set up a general theoretical framework and show that in a broad context, PCS fully recovers the support of $Ω$ and HCT-PCS yields optimal classification behavior. Our proofs shed interesting light on the behavior of stage-wise procedures.
Estimating High-dimensional Covariance and Precision Matrices under General Missing Dependence
完全观测数据的样本协方差阵$\boldsymbol{S}$是多种多元统计过程中的关键统计量,例如结构化协方差/精度矩阵估计、主成分分析和均值向量相等性检验。然而,当数据被部分观测时,来自可用数据的样本协方差阵存在偏差,并且不能提供有效的多变量程序。为了纠正偏差,一个简单的调整方法称为逆概率加权(IPW)已在以前的研究中使用,产生的IPW估计。估计量在缺失数据上下文中扮演$\boldsymbol{S}$的角色,因此可以将其插入现成的多变量程序。然而,IPW估计量的理论性质(例如浓度)仅在非常简单的缺失结构下建立;每个样本的每个变量都独立地以相等的概率缺失。研究了在一般缺失相依下部分观测时IPW估计量的偏差。我们证明了最佳收敛速度$O_p(\sqrt{\log p / n})$的IPW估计的基础上的元素明智的最大范数。我们也得到类似的偏差结果,即使当隐式假设(已知的平均值和/或缺失概率)放松。最优速率在估计精度矩阵时尤其重要,因为“元定理”声称IPW估计器的速率控制所得到的精度矩阵估计器的速率。在模拟研究中,我们讨论了IPW估计的非正半定性,并将该估计与具有实际意义的插补方法进行了比较。
Estimating Precision Matrices for High-Dimensional Interval-Valued Data
School of Economics and Management, Beihang University(北京航空航天大学经济管理学院);Key Laboratory of Complex System Analysis, Management and Decision (Beihang University)(复杂系统分析与决策重点实验室(北京航空航天大学));Advanced Institute of Information Technology, Peking University(北京大学信息科学技术学院前沿交叉研究院);Wangxuan Institute of Computer Technology, Peking University(北京大学王选计算机研究所)