Tensor Completion using Subspace Information
Abstract
Tensor completion has attracted significant attention in both applications and theoretical research. Under standard uniform sampling, existing polynomial-time guarantees generally require more observations than the number of degree of freedom, motivating the study of a possible statistical-to-computational gap in highly missing regimes. Fortunately, in many practical scenarios, side information is available, which can provide valuable insights to mitigate these challenges. In this paper, we introduce an algorithm called Tensor Completion using Subspace Information (TCSI) that incorporates side information through an estimated subspace. Our approach first extracts the subspace from the available side information and then reformulates tensor completion as a matrix regression problem. We provide a theoretical analysis showing that, when accurate subspace information is available, the required sample complexity is reduced to nearly linear order in the uncoupled ambient dimensions, removing the coupled-mode dimension from the leading term. Leveraging the estimated subspace information, we obtain a less stringent sufficient signal-to-noise ratio requirement than those in several existing passive-uniform-sampling guarantees. Under additional mild conditions, we obtain a sharper statistical error bound. Our theoretical findings are supported by numerical simulations. We apply TCSI to the reconstruction of global Total Electron Content (TEC) maps and observe lower reconstruction errors than the compared methods in our experiments.
Related
- Coupled Tensor-Tensor Completion Method with Applications in Drug Repurposing
- Optimal Personalized Subspace Learning for Multi-view Tensor Observations
- Tensor Orthogonal Subspace Split: Theory and Applications
- Robust low-rank tensor completion via factorized weighted tensor schatten-p norm minimization
- Information-Induced Training Geometry: Exact Reduction, Canonical Completion, and Structured Expressivity
BibTeX
@misc{li2026tensor,
title = {{Tensor Completion using Subspace Information}},
author = {Jingyang Li and Michael K. Ng},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.24501},
eprint = {2609.24501},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.24501},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Tensor Completion using Subspace Information
AU - Jingyang Li
AU - Michael K. Ng
PY - 2026
DA - 2026/09/21
PB - arXiv
DO - 10.48550/arxiv.2609.24501
UR - https://arxiv.org/abs/2609.24501
AB - Tensor completion has attracted significant attention in both applications and theoretical research. Under standard uniform sampling, existing polynomial-time guarantees generally require more observations than the number of degree of freedom, motivating the study of a possible statistical-to-computational gap in highly missing regimes. Fortunately, in many practical scenarios, side information is available, which can provide valuable insights to mitigate these challenges. In this paper, we introduce an algorithm called Tensor Completion using Subspace Information (TCSI) that incorporates side information through an estimated subspace. Our approach first extracts the subspace from the available side information and then reformulates tensor completion as a matrix regression problem. We provide a theoretical analysis showing that, when accurate subspace information is available, the required sample complexity is reduced to nearly linear order in the uncoupled ambient dimensions, removing the coupled-mode dimension from the leading term. Leveraging the estimated subspace information, we obtain a less stringent sufficient signal-to-noise ratio requirement than those in several existing passive-uniform-sampling guarantees. Under additional mild conditions, we obtain a sharper statistical error bound. Our theoretical findings are supported by numerical simulations. We apply TCSI to the reconstruction of global Total Electron Content (TEC) maps and observe lower reconstruction errors than the compared methods in our experiments.
ER -