Sigmoid Tensor Decomposition for Compact Representations of Binary Data
Abstract
Nonlinear low-rank decompositions are receiving growing attention as a way to represent structured data with fewer parameters than conventional linear methods. Existing nonlinear approaches have focused mainly on matrices, which limits their ability to capture multidimensional structure. We propose a nonlinear tensor decomposition based on sigmoid nonlinearization in the Tucker and tensor-train formats. A binary tensor is represented by a low-rank real-valued score tensor, whose parameters are estimated by minimizing the binary cross-entropy loss. The resulting optimization problem is jointly nonconvex, but becomes convex with respect to one factor or score block when the remaining factors are fixed, as in the corresponding matrix case. We evaluate the method on several structured binary tensors and compare it with conventional Tucker and tensor-train decompositions across a range of common ranks. These results indicate that sigmoid nonlinearization is particularly effective in the low-rank regime, where it can provide more accurate binary reconstructions than conventional linear tensor decompositions at comparable model sizes.
Related
- A Novel Sigmoid Fractional Binary Swarm Intelligent Solution for Optimal Reactive Power Dispatch
- Tensor products of 2-representations
- Eigensolvers for polynomial roots and tensor decomposition
- An irreducible decomposition of the Weil representation restricted to open compact subgroups
- Heteroskedastic Canonical Polyadic Tensor Decomposition
BibTeX
@misc{kopnina2026sigmoid,
title = {{Sigmoid Tensor Decomposition for Compact Representations of Binary Data}},
author = {Irina Kopnina and Sergey Matveev and Ivan Oseledets},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.08020},
eprint = {2611.08020},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.08020},
}
RIS
TY - UNPB
TI - Sigmoid Tensor Decomposition for Compact Representations of Binary Data
AU - Irina Kopnina
AU - Sergey Matveev
AU - Ivan Oseledets
PY - 2026
DA - 2026/11/05
PB - arXiv
DO - 10.48550/arxiv.2611.08020
UR - https://arxiv.org/abs/2611.08020
AB - Nonlinear low-rank decompositions are receiving growing attention as a way to represent structured data with fewer parameters than conventional linear methods. Existing nonlinear approaches have focused mainly on matrices, which limits their ability to capture multidimensional structure. We propose a nonlinear tensor decomposition based on sigmoid nonlinearization in the Tucker and tensor-train formats. A binary tensor is represented by a low-rank real-valued score tensor, whose parameters are estimated by minimizing the binary cross-entropy loss. The resulting optimization problem is jointly nonconvex, but becomes convex with respect to one factor or score block when the remaining factors are fixed, as in the corresponding matrix case. We evaluate the method on several structured binary tensors and compare it with conventional Tucker and tensor-train decompositions across a range of common ranks. These results indicate that sigmoid nonlinearization is particularly effective in the low-rank regime, where it can provide more accurate binary reconstructions than conventional linear tensor decompositions at comparable model sizes.
ER -