Detection of a sparse submatrix of a high-dimensional noisy matrix

Abstract

We observe a $N\times M$ matrix $Y_{ij}=s_{ij}+\xi_{ij}$ with $\xi_{ij}\sim {\mathcal {N}}(0,1)$ i.i.d. in $i,j$, and $s_{ij}\in \mathbb {R}$. We test the null hypothesis $s_{ij}=0$ for all $i,j$ against the alternative that there exists some submatrix of size $n\times m$ with significant elements in the sense that $s_{ij}\ge a>0$. We propose a test procedure and compute the asymptotical detection boundary $a$ so that the maximal testing risk tends to 0 as $M\to\infty$, $N\to\infty$, $p=n/N\to0$, $q=m/M\to0$. We prove that this boundary is asymptotically sharp minimax under some additional constraints. Relations with other testing problems are discussed. We propose a testing procedure which adapts to unknown $(n,m)$ within some given set and compute the adaptive sharp rates. The implementation of our test procedure on synthetic data shows excellent behavior for sparse, not necessarily squared matrices. We extend our sharp minimax results in different directions: first, to Gaussian matrices with unknown variance, next, to matrices of random variables having a distribution from an exponential family (non-Gaussian) and, finally, to a two-sided alternative for matrices with Gaussian elements.Comment: Published in at http://dx.doi.org/10.3150/12-BEJ470 the Bernoulli (http://isi.cbs.nl/bernoulli/) by the International Statistical Institute/Bernoulli Society (http://isi.cbs.nl/BS/bshome.htm

Adaptive Sensing for Estimation of Structured Sparse Signals

“…. , a stopping time T which indicates the time when we stop sampling, and the estimator S. In what follows we consider always the definition of the two last quantities given by (5) and (6). The queries Q t will be defined separately for each special case.…”

“…, n} : e i ∩ e T ′ = ∅ . The stopping time T and estimator S are defined as usual in (5) and (6). Note that this is an instance of the general procedure described in the setting of Proposition 3 with K = 1 and l 1 = 1.…”

“…The effect that structural information regarding the unknown parameter of interest has on statistical inference is an important question, and was studied in e.g. [2,1,6] in non-adaptive sensing settings, and investigated for a particular class of structural information in [3] in an adaptive sensing setting.…”

Optimal change point detection in Gaussian processes

“…, n}. CDEP as a risk measure has been considered in the literature for detecting abnormal clusters in a network (see e.g., [4,10]). It also provides an upper bound on the Bayesian risk measure.…”

Information limits for recovering a hidden community

“…This corresponds to a symmetric version of the submatrix localization problem studied in [37,30,10,9,31,12,11]. 1 When µ > 0, the entries of A with row and column indices in C * have positive mean µ except those on the diagonal, while the rest of the entries have zero mean.…”