Identifiability of Nonnegative Tucker Decompositions -- Part I: Theory
Not on the map. Why?
Abstract (arXiv)
Tensor decompositions have become a central tool in data science, with applications in areas such as data analysis, signal processing, and machine learning. A key property of many tensor decompositions, such as the canonical polyadic decomposition, is identifiability: the factors are unique, up to trivial scaling and permutation ambiguities. This allows one to recover the groundtruth sources that generated the data. The Tucker decomposition (TD) is a central and widely used tensor decomposition model. However, it is in general not identifiable. In this paper, we study the identifiability of the nonnegative TD (nTD). By adapting and extending identifiability results of nonnegative matrix factorization (NMF), we provide uniqueness results for nTD. Our results require the nonnegative matrix factors to have some degree of sparsity (namely, satisfy the separability condition, or the sufficiently scattered condition), while the core tensor only needs to have some slices (or linear combinations of them) or unfoldings with full column rank (but does not need to be nonnegative). Under such conditions, we derive several procedures, using either unfoldings or slices of the input tensor, to obtain identifiable nTDs by minimizing the volume of unfoldings or slices of the core tensor.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 22 of its 55 references that are in the map. First posted on arXiv 19 May 2025.
Builds on
22 of 55 references.
References · 22
- 2016Tensor Decomposition for Signal Processing and Machine Learningcited by 223
- 2016TensorLy: Tensor Learning in Pythoncited by 104
- 2018Nonnegative Matrix Factorization for Signal and Data Analytics: Identifiability, Algorithms, and Applicationscited by 55
- 2017On Identifiability of Nonnegative Matrix Factorizationcited by 34
- 2015A Flexible and Efficient Algorithmic Framework for Constrained Matrix and Tensor Factorizationcited by 30
- 2016Robust Volume Minimization-Based Matrix Factorization for Remote Sensing and Document Clusteringcited by 28
- 2014Identifiability of the Simplex Volume Minimization Criterion for Blind Hyperspectral Unmixing: The No Pure-Pixel Casecited by 26
- 2020Simplex-Structured Matrix Factorization: Sparsity-based Identifiability and Provably Correct Algorithmscited by 21
- 2014Efficient Nonnegative Tucker Decompositions: Algorithms and Uniquenesscited by 17
- 2019Multiway clustering via tensor block modelscited by 17
- 2019Blind Audio Source Separation with Minimum-Volume Beta-Divergence NMFcited by 16
- 2016Anchor-Free Correlated Topic Modeling: Identifiability and Algorithmcited by 15
10 more
- 2020Computing Large-Scale Matrix and Tensor Decomposition with Structured Factors: A Unified Nonconvex Optimization Perspectivecited by 14
- 2021Uncovering audio patterns in music with Nonnegative Tucker Decomposition for structural segmentationcited by 8
- 2022Estimating Higher-Order Mixed Memberships via the ℓ2,∞ Tensor Perturbation Boundcited by 8
- 2013When are Overcomplete Topic Models Identifiable? Uniqueness of Tensor Tucker Decompositions with Structured Sparsitycited by 7
- 2021HyperNTF: A Hypergraph Regularized Nonnegative Tensor Factorization for Dimensionality Reductioncited by 5
- 2024Checking the Sufficiently Scattered Condition using a Global Non-Convex Optimization Softwarecited by 3
- 2022Complete Decomposition of Symmetric Tensors in Linear Time and Polylogarithmic Precisioncited by 2
- 2024An efficient uniqueness theorem for overcomplete tensor decompositioncited by 2
- 2024Undercomplete Decomposition of Symmetric Tensors in Linear Time, and Smoothed Analysis of the Condition Numbercited by 1
- 2024Low-Rank Matrix Factorizations with Volume-based Constraints and Regularizationscited by 1
Built on by
Citing papers per year: 2025: 1.
1 paper.