Minimax bounds for sparse PCA with noisy high-dimensional data

DOI10.1214/12-AOS1014zbMATH Open1292.62071arXiv1203.0967WikidataQ30862008 ScholiaQ30862008MaRDI QIDQ366956FDOQ366956

Authors: Aharon Birnbaum, Iain M. Johnstone, Boaz Nadler, Debashis Paul 

Publication date: 25 September 2013

Published in: The Annals of Statistics (Search for Journal in Brave)

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.

Full work available at URL: https://arxiv.org/abs/1203.0967

Recommendations

• Minimax sparse principal subspace estimation in high dimensions
• Sparse PCA: optimal rates and adaptive estimation
• Sparse PCA via covariance thresholding
• Statistical and computational trade-offs in estimation of sparse principal components
• Sparse principal component analysis and iterative thresholding

zbMATH Keywords

high-dimensional dataprincipal component analysissparsityminimax riskspiked covariance model

Mathematics Subject Classification ID

Asymptotic properties of nonparametric inference (62G20) Factor analysis and principal components; correspondence analysis (62H25)

Cites Work

• Principal component analysis.
• Title not available (Why is that?)
• Covariance regularization by thresholding
• Regularized estimation of large covariance matrices
• Eigenvalues of large sample covariance matrices of spiked population models
• On consistency and sparsity for principal components analysis in high dimensions
• Asymptotics of sample eigenstructure for a large dimensional spiked covariance model
• A Direct Formulation for Sparse PCA Using Semidefinite Programming
• Finite sample approximation results for principal component analysis: A matrix perturbation approach
• Consistency of sparse PCA in high dimension, low sample size contexts
• Local operator theory, random matrices and Banach spaces.
• Probabilistic Principal Component Analysis
• Sparse principal component analysis via regularized low rank matrix approximation
• Operator norm consistent estimation of large-dimensional sparse covariance matrices
• Sparse PCA: optimal rates and adaptive estimation
• Adaptive thresholding for sparse covariance matrix estimation
• Generalized thresholding of large covariance matrices
• Optimal rates of convergence for covariance matrix estimation
• Sphere packings. (Edited by John Talbot)
• Title not available (Why is that?)
• Asymptotic Theory for Principal Component Analysis
• Information-theoretic determination of minimax rates of convergence
• Discussion
• High-dimensional analysis of semidefinite relaxations for sparse principal components
• Chi-square oracle inequalities
• Winding number criterion for existence and uniqueness of equilibrium in linear rational expectations models

Cited In (61)

• Exactly Uncorrelated Sparse Principal Component Analysis
• Long random matrices and tensor unfolding
• Quantitative limit theorems and bootstrap approximations for empirical spectral projectors
• High-Dimensional Dynamic Covariance Matrices With Homogeneous Structure
• A Synthetic Regression Model for Large Portfolio Allocation
• Lower bound estimation for a family of high-dimensional sparse covariance matrices
• Large sample correlation matrices with unbounded spectrum
• Noisy low-rank matrix completion under general bases
• Sparse constrained projection approximation subspace tracking
• Sparse principal component analysis for high‐dimensional stationary time series
• Limiting laws for divergent spiked eigenvalues and largest nonspiked eigenvalue of sample covariance matrices
• Title not available (Why is that?)
• Robust covariance estimation for approximate factor models
• Do semidefinite relaxations solve sparse PCA up to the information limit?
• Majorization-Minimization on the Stiefel Manifold With Application to Robust Sparse PCA
• Recovery of simultaneous low rank and two-way sparse coefficient matrices, a nonconvex approach
• On two-sample mean tests under spiked covariances
• Lower bounds for sparse coding
• A simultaneous test of mean vector and covariance matrix in high-dimensional settings
• Influence diagnostics in support vector machines
• Detection of hubs in complex networks by the Laplacian matrix
• Testing and estimating change-points in the covariance matrix of a high-dimensional time series
• Optimal estimation and rank detection for sparse spiked covariance matrices
• Principal component analysis: a review and recent developments
• On the optimality of sliced inverse regression in high dimensions
• Sparse PCA: optimal rates and adaptive estimation
• The spectral norm of random inner-product kernel matrices
• Sparse PCA via covariance thresholding
• Random matrix theory in statistics: a review
• Estimating structured high-dimensional covariance and precision matrices: optimal rates and adaptive estimation
• Inference for low-rank models
• Lower bounds for invariant statistical models with applications to principal component analysis
• Minimax estimation in sparse canonical correlation analysis
• Asymptotically efficient estimation of smooth functionals of covariance operators
• Recent developments in high dimensional covariance estimation and its related issues, a review
• Testing the order of a population spectral distribution for high-dimensional data
• Statistical inference for principal components of spiked covariance matrices
• Estimation of functionals of sparse covariance matrices
• New asymptotic results in principal component analysis
• Optimal sparse eigenspace and low-rank density matrix estimation for quantum systems
• Bayesian inference for spectral projectors of the covariance matrix
• An asymptotically minimax kernel machine
• Large covariance estimation through elliptical factor models
• Random matrix theory and its applications
• ROP: matrix recovery via rank-one projections
• Sparse equisigned PCA: algorithms and performance bounds in the noisy rank-1 setting
• Detecting changes in the second moment structure of high-dimensional sensor-type data in a \(K\)-sample setting
• A high-dimensional test on linear hypothesis of means under a low-dimensional factor model
• Sparse PCA-based on high-dimensional Itô processes with measurement errors
• Large Covariance Estimation by Thresholding Principal Orthogonal Complements
• Rate-optimal posterior contraction for sparse PCA
• High-probability bounds for the reconstruction error of PCA
• Optimal detection of sparse principal components in high dimension
• Robust covariance and scatter matrix estimation under Huber's contamination model
• Posterior contraction rates of the phylogenetic Indian buffet processes
• Testing equivalence of clustering
• Rejoinder of ``Estimating structured high-dimensional covariance and precision matrices: optimal rates and adaptive estimation
• The limits of the sample spiked eigenvalues for a high-dimensional generalized Fisher matrix and its applications
• Sparsistency and agnostic inference in sparse PCA
• Optimality and sub-optimality of PCA. I: Spiked random matrix models
• Minimax sparse principal subspace estimation in high dimensions

Uses Software

• PMA

This page was built for publication: Minimax bounds for sparse PCA with noisy high-dimensional data

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q366956)