Abstract:SIAM Journal on Matrix Analysis and Applications, Volume 45, Issue 4, Page 1805-1826, December 2024. Nonnegative matrix factorization (NMF) is an important unsupervised learning method to extract meaningful features from data. To address the NMF problem within a polynomial time framework, researchers have introduced a separability assumption, which has recently evolved into the concept of coseparability. This advancement offers a more efficient core representation for the original data. However, in the real world, the data is more naturally represented as a multidimensional array, such as images or videos. The NMF's application to high-dimensional data involves vectorization, which risks losing essential multidimensional correlations. To retain these inherent correlations in the data, we turn to tensors (multidimensional arrays) and leverage the tensor t-product. This approach extends the coseparable NMF to the tensor setting, creating what we term coseparable nonnegative tensor factorization (NTF). In this work, we provide an alternating index selection method to select the coseparable core. Furthermore, we validate the t-CUR sampling theory and integrate it with the tensor discrete empirical interpolation method to introduce an alternative, randomized index selection process. These methods have been tested on both synthetic and facial analysis datasets. The results demonstrate the efficiency of coseparable NTF when compared to coseparable NMF.

Nonnegative Tensor Factorizations Using an Alternating Direction Method

Multiplicative Algorithms for Symmetric Nonnegative Tensor Factorizations and Its Applications.

Non-negative tensor factorization with applications to statistics and computer vision

An Alternating Direction Algorithm for Matrix Completion with Nonnegative Factors

Nonnegative Tensor Cofactorization and Its Unified Solution

Non-negative Tri-factor Tensor Decomposition with Applications

Coseparable Nonnegative Tensor Factorization with t-CUR Decomposition

Learning Nonnegative Factors from Tensor Data: Probabilistic Modeling and Inference Algorithm

Nonnegative Tensor Factorization, Completely Positive Tensors and an Hierarchical Elimination Algorithm

Nonnegative Tensor Factorization, Completely Positive Tensors, and a Hierarchical Elimination Algorithm.

Numerical Method for the Generalized Nonnegative Tensor Factorization Problem

Unified Solution to Nonnegative Data Factorization Problems

Adversarially-Trained Nonnegative Matrix Factorization

Nonnegative Low Rank Tensor Approximation and its Application to Multi-dimensional Images

Efficient Constrained Tensor Factorization by Alternating Optimization with Primal-Dual Splitting

A Fast Algorithm for Nonnegative Matrix Factorization and Its Convergence.

Non-negative matrix factorization with α-divergence

Convex and semi-nonnegative matrix factorizations

Modified Subspace Barzilai-Borwein Gradient Method for Non-Negative Matrix Factorization

Noisy Nonnegative Tucker Decomposition with Sparse Factors and Missing Data