ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 1109.0898
168
125

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

5 September 2011
C. Butucea
Yu. I. Ingster
ArXivPDFHTML
Abstract

We observe a N×MN\times MN×M matrix Yij=sij+ξijY_{ij}=s_{ij}+\xi_{ij}Yij​=sij​+ξij​ with ξij∼N(0,1)\xi_{ij}\sim {\mathcal {N}}(0,1)ξij​∼N(0,1) i.i.d. in i,ji,ji,j, and sij∈Rs_{ij}\in \mathbb {R}sij​∈R. We test the null hypothesis sij=0s_{ij}=0sij​=0 for all i,ji,ji,j against the alternative that there exists some submatrix of size n×mn\times mn×m with significant elements in the sense that sij≥a>0s_{ij}\ge a>0sij​≥a>0. We propose a test procedure and compute the asymptotical detection boundary aaa so that the maximal testing risk tends to 0 as M→∞M\to\inftyM→∞, N→∞N\to\inftyN→∞, p=n/N→0p=n/N\to0p=n/N→0, q=m/M→0q=m/M\to0q=m/M→0. 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)(n,m)(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.

View on arXiv
Comments on this paper