Abstract:Real-world datasets often contain missing or corrupted values. Completing multidimensional tensor-structured data with missing entries is essential for numerous applications. Smoothness-constrained low-rank factorization models have shown superior performance with reduced computational costs. While effective at capturing global and long-range correlations, these models struggle to reproduce short-scale, high-frequency variations in the data. In this paper, we introduce the \Generalized Least Squares Kernelized Tensor Factorization (GLSKF) framework for tensor completion. GLSKF integrates smoothness-constrained low-rank factorization with a locally correlated residual process; the resulting additive structure can effectively characterize both global dependencies and local variations. In particular, we define the covariance norm to enforce the smoothness of factor matrices in the global low-rank factorization, and use structured covariance/kernel functions to model the local processes. For model estimation, we develop an alternating least squares (ALS) procedure with closed-form solutions for each subproblem. To efficiently handle missing data, GLSKF utilizes projection matrices that preserve the Kronecker structure of covariances, facilitating fast computations through conjugate gradient (CG) and preconditioned conjugate gradient (PCG) algorithms. The proposed framework is evaluated on four real-world datasets across diverse tasks: traffic speed imputation, color image inpainting, video completion, and MRI image reconstruction. Experimental results confirm that GLSKF delivers superior effectiveness and scalability, establishing it as a robust solution for multidimensional tensor completion.

Factor-driven completion of tensor data with missing entries

A Biased Deep Tensor Factorization Network For Tensor Completion

T-product factorization based method for matrix and tensor completion problems

Completion of High Order Tensor Data with Missing Entries via Tensor-train Decomposition

Tensor Time Series Imputation through Tensor Factor Modelling

Tensor Factorization for Low-Rank Tensor Completion.

MLCTR: A Fast Scalable Coupled Tensor Completion Based on Multi-Layer Non-Linear Matrix Factorization

Imbalanced low-rank tensor completion via latent matrix factorization

Tucker factorization-based tensor completion for robust traffic data imputation

Generalized Least Squares Kernelized Tensor Factorization

Tensor completion using total variation and low-rank matrix factorization

Higher-dimension Tensor Completion via Low-rank Tensor Ring Decomposition

Robust Low-tubal-rank Tensor Completion based on Tensor Factorization and Maximum Correntopy Criterion

Novel Factorization Strategies for Higher Order Tensors: Implications for Compression and Recovery of Multi-linear Data

Time-Aware Tensor Decomposition for Missing Entry Prediction

Exponential Family Tensor Completion with Auxiliary Information

Low Multilinear Rank Tensor Completion with Missing Data

Censored Least Squares for Imputing Missing Values in PARAFAC Tensor Factorization

Low-Rank Tensor Completion Using Matrix Factorization Based on Tensor Train Rank and Total Variation

Noisy Nonnegative Tucker Decomposition with Sparse Factors and Missing Data

A Novel Truncated Normal Tensor Completion Method for Multi-Source Fusion Data