Minimax bounds for sparse PCA with noisy high-dimensional data
Authors:
Preprint
Abstract
We study the problem of estimating the leading eigenvectors of a high-dimensional population covariance matrix based on independent Gaussian observations. We establish a lower bound on the minimax risk of estimators under the $l_2$ loss, in the joint limit as dimension and sample size increase to infinity, under various models of sparsity for the population eigenvectors. The lower bound on the risk points to the existence of different regimes of sparsity of the eigenvectors. We also propose a new method for estimating the eigenvectors by a two-stage coordinate selection scheme.
Keywords
All Related Versions
This publication has 20 references indexed in Scilit:
- Consistency of sparse PCA in High Dimension, Low Sample Size contextsJournal of Multivariate Analysis, 2013
- Adaptive Thresholding for Sparse Covariance Matrix EstimationJournal of the American Statistical Association, 2011
- Determining the Number of Factors from Empirical Distribution of EigenvaluesReview of Economics and Statistics, 2010
- On Consistency and Sparsity for Principal Components Analysis in High DimensionsJournal of the American Statistical Association, 2009
- A penalized matrix decomposition, with applications to sparse principal components and canonical correlation analysisBiostatistics, 2009
- Generalized Thresholding of Large Covariance MatricesJournal of the American Statistical Association, 2009
- Determining the number of components in a factor model from limited noisy dataChemometrics and Intelligent Laboratory Systems, 2008
- Regularized estimation of large covariance matricesThe Annals of Statistics, 2008
- A Direct Formulation for Sparse PCA Using Semidefinite ProgrammingSIAM Review, 2007
- Eigenvalues of large sample covariance matrices of spiked population modelsJournal of Multivariate Analysis, 2006