Abstract: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 natural to be represented as a multi-dimensional array, such as images or videos. The NMF's application to high-dimensional data involves vectorization, which risks losing essential multi-dimensional 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 (t-DEIM) 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 Tucker Decomposition with Beta-divergence for Music Structure Analysis of Audio Signals

Uncovering audio patterns in music with Nonnegative Tucker Decomposition for structural segmentation

Multiplicative Algorithms for Symmetric Nonnegative Tensor Factorizations and Its Applications.

Deep Nonnegative Matrix Factorization with Beta Divergences

Orthogonal Nonnegative Tucker Decomposition

Multifactor Sparse Feature Extraction Using Convolutive Nonnegative Tucker Decomposition

Noisy Nonnegative Tucker Decomposition with Sparse Factors and Missing Data

Online algorithms for Nonnegative Matrix Factorization with the Itakura-Saito divergence

Alternating proximal gradient method for sparse nonnegative Tucker decomposition

VOLUME-REGULARIZED NONNEGATIVE TUCKER DECOMPOSITION WITH IDENTIFIABILITY GUARANTEES

Non-negative matrix factorization with α-divergence

Non-negative Tensor Factorization for Speech Enhancement

An Iterative Reweighted Method for Tucker Decomposition of Incomplete Multiway Tensors

Perturbations of the Tcur Decomposition for Tensor Valued Data in the Tucker Format

Automatic Relevance Determination in Nonnegative Matrix Factorization with the β-Divergence

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

Non-negative Tensor Mixture Learning for Discrete Density Estimation

Supervised Nonnegative Tucker Decomposition for Computational Phenotyping.

Algorithms for Non-Negative Matrix Factorization on Noisy Data With Negative Values

Coseparable Nonnegative Tensor Factorization With T-CUR Decomposition

Multiplex Transformed Tensor Decomposition for Multidimensional Image Recovery.