A super-resolution framework for tensor decomposition
Authors:
- 6 April 2022
- journal article
- research article
- Published by Oxford University Press (OUP) in Information and Inference: A Journal of the IMA
- Vol. 11 (4) , 1287-1328
Abstract
This work considers a super-resolution framework forovercomplete tensor decomposition. Specifically, we view tensor decomposition as a super-resolution problem of recovering a sum of Dirac measures on the sphere and solve it by minimizing a continuous analog of the $\ell _1$ norm on the space of measures. The optimal value of this optimization defines the tensor nuclear norm. Similar to the separation condition in the super-resolution problem, by explicitly constructing a dual certificate, we develop incoherence conditions of the tensor factors so that they form the unique optimal solution of the continuous analog of $\ell _1$ norm minimization. Remarkably, the derived incoherence conditions are satisfied with high probability by random tensor factors uniformly distributed on the sphere, implying global identifiability of random tensor factors.
Keywords
All Related Versions
Funding Information
- National Science Foundation (DMS-1913039, CCF-2203060, CCF-2106834)
This publication has 38 references indexed in Scilit:
- Nuclear norm of higher-order tensorsMathematics of Computation, 2017
- On Tensor Completion via Nuclear Norm MinimizationFoundations of Computational Mathematics, 2015
- The Convex Geometry of Linear Inverse ProblemsFoundations of Computational Mathematics, 2012
- Tensor completion and low-n-rank tensor recovery via convex optimizationInverse Problems, 2011
- Exact Matrix Completion via Convex OptimizationFoundations of Computational Mathematics, 2009
- The restricted isometry property and its implications for compressed sensingComptes Rendus Mathématique, 2008
- Optimally sparse representation in general (nonorthogonal) dictionaries via ℓ 1 minimizationProceedings of the National Academy of Sciences, 2003
- A nonlinear programming algorithm for solving semidefinite programs via low-rank factorizationMathematical Programming, 2003
- Characterization of the subdifferential of some matrix normsLinear Algebra and its Applications, 1992
- Three-way arrays: rank and uniqueness of trilinear decompositions, with application to arithmetic complexity and statisticsLinear Algebra and its Applications, 1977