Rotational Invariant Estimator for General Noisy Matrices

DOI10.1109/TIT.2016.2616132MaRDI QIDQ2976467zbMATH OpenFDO

Authors Joël Bun, Romain Allez, Jean-Philippe Bouchaud, Marc Potters

Publication date 28 April 2017

Published in IEEE Transactions on Information Theory (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1502.06736

Mathematics Subject Classification ID

Applications of statistics in engineering and industry; control charts (62P30) Signal theory (characterization, reconstruction, filtering, etc.) (94A12)

Abstract: We investigate the problem of estimating a given real symmetric signal matrix $extbfC$ from a noisy observation matrix $extbfM$ in the limit of large dimension. We consider the case where the noisy measurement $extbfM$ comes either from an arbitrary additive or multiplicative rotational invariant perturbation. We establish, using the Replica method, the asymptotic global law estimate for three general classes of noisy matrices, significantly extending previously obtained results. We give exact results concerning the asymptotic deviations (called overlaps) of the perturbed eigenvectors away from the true ones, and we explain how to use these overlaps to "clean" the noisy eigenvalues of $extbfM$. We provide some numerical checks for the different estimators proposed in this paper and we also make the connection with some well known results of Bayesian statistics.

Cited in

(28)

• Eigenvectors distribution and quantum unique ergodicity for deformed Wigner matrices
• The Harish-Chandra integral: an introduction with examples
• Approximate message passing algorithms for rotationally invariant matrices
• Improving portfolios global performance using a cleaned and robust covariance matrix estimate
• Convergence of eigenvector empirical spectral distribution of sample covariance matrices
• High dimensional deformed rectangular matrices with applications in matrix denoising
• Empirical spectral distribution of a matrix under perturbation
• PCA meets RG
• Statistical diagonalization of a random biased Hamiltonian: the case of the eigenvectors
• Composite estimator based on the recursive ratio for an arbitrary rotation scheme
• Perturbative construction of mean-field equations in extensive-rank matrix factorization and denoising
• A review of two decades of correlations, hierarchies, networks and clustering in financial markets
• Cleaning large correlation matrices: tools from random matrix theory
• Two short pieces around the Wigner problem
• Filtering time-dependent covariance matrices using time-independent eigenvalues
• A Compound Decision Approach to Covariance Matrix Estimation
• Curvature-driven pathways interpolating between stationary points: the case of the pure spherical 3-spin model
• Local laws for multiplication of random matrices
• Spiked multiplicative random matrices and principal components
• The decimation scheme for symmetric matrix factorization
• Freeness of type B and conditional freeness for random matrices
• Matrix denoising: Bayes-optimal estimators via low-degree polynomials
• Optimality of approximate message passing for spiked matrix models with rotationally invariant noise
• Rectangular rotational invariant estimator for high-rank matrix estimation
• Eigenvector overlaps of random covariance matrices and their submatrices
• Blind free deconvolution over one-parameter sparse families via eigenmatrix
• The Wishart-Rosenzweig-Porter random matrix ensemble
• Scaling and renormalization in high-dimensional regression

This page was built for publication: Rotational Invariant Estimator for General Noisy Matrices

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2976467)