Learnable Transform-Assisted Tensor Decomposition for Spatio-Irregular Multidimensional Data Recovery

Hao Zhang, Ting Zhu Huang*, Xi Le Zhao*, Shuqin Zhang, Jin Yu Xie, Tai Xiang Jiang, Michael K. Ng

*Corresponding author for this work

Research output: Contribution to journalJournal articlepeer-review

Abstract

Tensor decompositions have been successfully applied to multidimensional data recovery. However, classical tensor decompositions are not suitable for emerging spatio-irregular multidimensional data (i.e., spatio-irregular tensor), whose spatial domain is non-rectangular, e.g., spatial transcriptomics data from bioinformatics and semantic units from computer vision. By using preprocessing (e.g., zero-padding or element-wise 0-1 weighting), the spatio-irregular tensor can be converted to a spatio-regular tensor and then classical tensor decompositions can be applied, but this strategy inevitably introduces bias information, leading to artifacts. How to design a tensor-based method suitable for emerging spatio-irregular tensors is an imperative challenge. To address this challenge, we propose a learnable transform-assisted tensor singular value decomposition (LTA-TSVD) for spatio-irregular tensor recovery, which allows us to leverage the intrinsic structure behind the spatio-irregular tensor. Specifically, we design a learnable transform to project the original spatio-irregular tensor into its latent spatio-regular tensor, and then the latent low-rank structure is captured by classical TSVD on the resulting regular tensor. Empowered by LTA-TSVD, we develop spatio-irregular low-rank tensor completion (SIR-LRTC) and spatio-irregular tensor robust principal component analysis (SIR-TRPCA) models for the spatio-irregular tensor imputation and denoising respectively, and we design corresponding solving algorithms with theoretical convergence. Extensive experiments including the spatial transcriptomics data imputation and hyperspectral image denoising show SIR-LRTC and SIR-TRPCA are superior performance to competing approaches and benefit downstream applications.

Original languageEnglish
Article number12
Pages (from-to)1-23
Number of pages23
JournalACM Transactions on Knowledge Discovery from Data
Volume19
Issue number1
Early online date11 Dec 2024
DOIs
Publication statusPublished - Jan 2025

Scopus Subject Areas

  • General Computer Science

User-Defined Keywords

  • low-rank tensor
  • proximal alternating minimization
  • Spatio-irregular tensor
  • tensor learning

Fingerprint

Dive into the research topics of 'Learnable Transform-Assisted Tensor Decomposition for Spatio-Irregular Multidimensional Data Recovery'. Together they form a unique fingerprint.

Cite this