Sharp detection of smooth signals in a high-dimensional sparse matrix with indirect observations

DOI10.1214/15-AIHP689MaRDI QIDQ503078zbMATH OpenOpenAlexFDO

Authors Cristina Butucea, Ghislaine Gayraud

Publication date 11 January 2017

Published in Annales de l'Institut Henri Poincaré. Probabilités et Statistiques (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1301.4660

zbMATH Keywords

high-dimensional datasparsityinverse problemsdetection boundaryGaussian white noise modelasymptotic minimax testheterogeneous observationsindirect observationssharp rates

Mathematics Subject Classification ID

Nonparametric hypothesis testing (62G10) Asymptotic properties of nonparametric inference (62G20) Gaussian processes (60G15) Hypothesis testing in multivariate analysis (62H15) Minimax procedures in statistical decision theory (62C20)

Abstract: We consider a matrix-valued Gaussian sequence model, that is, we observe a sequence of high-dimensional $MimesN$ matrices of heterogeneous Gaussian random variables $x_{ij,k}$ for $iin1,\mathrm{...},M$, $jin1,\mathrm{...},N$ and $kinmathbbZ$. The standard deviation of our observations is $ep{k}^s$ for some $ep>0$ and $sgeq0$. We give sharp rates for the detection of a sparse submatrix of size $mimesn$ with active components. A component $(i,j)$ is said active if the sequence ${x_{ij,k}}_k$ have mean ${het{a}_{ij,k}}_k$ within a Sobolev ellipsoid of smoothness $au>0$ and total energy $su{m}_{k}het{a}_{ij,k}^2$ larger than some $r_e^{2}p$. Our rates involve relationships between $m,,n,,M$ and $N$ tending to infinity such that $m/M$, $n/N$ and $ep$ tend to 0, such that a test procedure that we construct has asymptotic minimax risk tending to 0. We prove corresponding lower bounds under additional assumptions on the relative size of the submatrix in the large matrix of observations. Except for these additional conditions our rates are asymptotically sharp. Lower bounds for hypothesis testing problems mean that no test procedure can distinguish between the null hypothesis (no signal) and the alternative, i.e. the minimax risk for testing tends to 1.

Recommendations

• Signal detection in high dimension: the multispiked case
• Detection of a sparse submatrix of a high-dimensional noisy matrix
• Sharp optimality for high-dimensional covariance testing under sparse signals
• Optimal signal detection in some spiked random matrix models: likelihood ratio tests and linear spectral statistics
• Statistical and computational limits for sparse matrix detection
• Optimal detection of sparse principal components in high dimension
• Detecting weak signals in high dimensions
• Robust test for detecting a signal in a high dimensional sparse normal vector
• Optimal detection of multi-sample aligned sparse signals
• Detection thresholds in very sparse matrix completion

Cited in

(8)

• Detecting weak signals in high dimensions
• Sharp minimax tests for large covariance matrices and adaptation
• Sharp detection in PCA under correlations: all eigenvalues matter
• Detection of a sparse submatrix of a high-dimensional noisy matrix
• Distribution-free, size adaptive submatrix detection with acceleration
• Signal detection in high dimension: the multispiked case
• Connection between the selection problem for a sparse submatrix of a large-size matrix and the Bayesian problem of hypotheses testing
• Sharp variable selection of a sparse submatrix in a high-dimensional noisy matrix

This page was built for publication: Sharp detection of smooth signals in a high-dimensional sparse matrix with indirect observations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q503078)