A generalized least-square matrix decomposition

DOI10.1080/01621459.2013.852978zbMATH Open1367.62184arXiv1102.3074OpenAlexW2030854293MaRDI QIDQ4975339FDOQ4975339

Authors: Genevera I. Allen, Logan Grosenick, Jonathan Taylor 

Publication date: 4 August 2017

Published in: Journal of the American Statistical Association (Search for Journal in Brave)

Abstract: Variables in many massive high-dimensional data sets are structured, arising for example from measurements on a regular grid as in imaging and time series or from spatial-temporal measurements as in climate studies. Classical multivariate techniques ignore these structural relationships often resulting in poor performance. We propose a generalization of the singular value decomposition (SVD) and principal components analysis (PCA) that is appropriate for massive data sets with structured variables or known two-way dependencies. By finding the best low rank approximation of the data with respect to a transposable quadratic norm, our decomposition, entitled the Generalized least squares Matrix Decomposition (GMD), directly accounts for structural relationships. As many variables in high-dimensional settings are often irrelevant or noisy, we also regularize our matrix decomposition by adding two-way penalties to encourage sparsity or smoothness. We develop fast computational algorithms using our methods to perform generalized PCA (GPCA), sparse GPCA, and functional GPCA on massive data sets. Through simulations and a whole brain functional MRI example we demonstrate the utility of our methodology for dimension reduction, signal recovery, and feature selection with high-dimensional structured data.

Full work available at URL: https://arxiv.org/abs/1102.3074

Recommendations

• A penalized matrix decomposition, with applications to sparse principal components and canonical correlation analysis
• A generalization of principal component analysis to \(K\) sets of variables.
• On consistency and sparsity for principal components analysis in high dimensions
• Sparse-smooth regularized singular value decomposition
• Sparse principal component analysis in Hilbert space

zbMATH Keywords

singular value decompositionsparse PCAfunctional PCAprincipal components analysisneuroimagingstructured data

Mathematics Subject Classification ID

Factor analysis and principal components; correspondence analysis (62H25) Applications of statistics to biology and medical sciences; meta analysis (62P10) Image analysis in multivariate analysis (62H35)

Cites Work

• Bi-cross-validation of the SVD and the nonnegative matrix factorization
• SparseNet: coordinate descent with nonconvex penalties
• The analysis of two-way functional data using two-way regularized singular value decompositions
• Matrix Analysis
• Variable Selection via Nonconcave Penalized Likelihood and its Oracle Properties
• Title not available (Why is that?)
• Sparsity and Smoothness Via the Fused Lasso
• Model Selection and Estimation in Regression with Grouped Variables
• The solution path of the generalized lasso
• On consistency and sparsity for principal components analysis in high dimensions
• Local multidimensional scaling for nonlinear dimension reduction, graph drawing, and proximity analysis
• Sparse principal component analysis via regularized low rank matrix approximation
• Smoothed functional principal components analysis by choice of norm
• On the ``degrees of freedom of the lasso
• A likelihood ratio test for separability of covariances
• The statistical analysis of fMRI data
• Transposable regularized covariance models with an application to missing data imputation
• On the degrees of freedom in shrinkage estimation
• Laplacian matrices of graphs: A survey
• Biclustering via sparse singular value decomposition
• The statistical analysis of functional MRI data
• On the inverses of some patterned matrices arising in the theory of stationary time series
• Learning to decode cognitive states from brain images
• Inference on low-rank data matrices with applications to microarray data
• Multivariate data analysis: The French way

Cited In (28)

• PLS for Big Data: a unified parallel algorithm for regularised group PLS
• Generalized matrix decomposition regression: estimation and inference for two-way structured data
• Heteroscedastic BART via Multiplicative Regression Trees
• Estimating a common period for a set of irregularly sampled functions with applications to periodic variable star data
• Search for weights in the problem of finite-rank signal estimation in the presence of random noise
• Mathematical characterization of sophisticated variants for relevance learning in learning matrix quantization based on Schatten-\(p\)-norms
• Discussion of the paper ‘A general framework for functional regression modelling’
• SVD-based incremental approaches for recommender systems
• Deconvolution of a discrete uniform distribution
• Flow-directed PCA for monitoring networks
• Principal component analysis for river network data: use of spatiotemporal correlation and heterogeneous covariance structure
• Matrix factorization for multivariate time series analysis
• A Bayesian Partially Observable Online Change Detection Approach with Thompson Sampling
• Accounting for time dependence in large-scale multiple testing of event-related potential data
• An oracle inequality for quasi-Bayesian nonnegative matrix factorization
• Functional factor analysis for periodic remote sensing data
• Statistical Foundations Driving 21st Century Innovation
• Local-aggregate modeling for big data via distributed optimization: applications to neuroimaging
• Application of the sequential matrix diagonalization algorithm to high-dimensional functional MRI data
• Weighted norms in subspace-based methods for time series analysis.
• Local biplots for multidimensional scaling, with application to the microbiome
• Multilevel Matrix-Variate Analysis and its Application to Accelerometry-Measured Physical Activity in Clinical Populations
• Multiple correspondence analysis and the multilogit bilinear model
• Deviance matrix factorization
• Population value decomposition, a framework for the analysis of image populations
• Interpretability of composite indicators based on principal components
• Adaptive gPCA: a method for structured dimensionality reduction with applications to microbiome data
• Blind deconvolution of covariance matrix inverses for autoregressive processes

This page was built for publication: A generalized least-square matrix decomposition

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4975339)