What problem does this paper attempt to address?

The problem that this paper attempts to solve is related to the convergence of the Alternating Least - Squares (ALS) in tensor decomposition. Specifically, the paper mainly focuses on the convergence analysis of the Regularized Alternating Least - Squares (RALS) method to improve the performance of ALS when dealing with non - degenerate swamps. ### Problem Background 1. **Limitations of ALS**: - ALS is a commonly used CANDECOMP/PARAFAC (CP) decomposition method, but it has some drawbacks. For example, initialization problems, collinearity or degeneracy of factor matrices may lead to a slow convergence speed, resulting in the so - called "swamp" phenomenon. - Especially in non - degenerate cases, ALS may encounter the problem of slow convergence, which is manifested as a flat curve in the log - error plot. 2. **Introduction of RALS**: - RALS is an improved version of ALS. By introducing a regularization term, it can accelerate convergence and reduce the influence of the swamp phenomenon. - The cost function of RALS penalizes the difference between the factor matrices of the current and previous iterations, thus avoiding the slow convergence problem that may occur in ALS. ### Research Objectives The main objectives of the paper are: - **Analyze the convergence of RALS**: Prove that the limit point of the convergent sequence of the RALS algorithm is a critical point of the original ALS minimization problem. - **Compare the performance of ALS and RALS**: Demonstrate the superiority of RALS in dealing with non - degenerate swamps through numerical experiments, especially when the initial factor selection is not good. ### Main Results - **Theoretical Analysis**: The paper proves that if the sequence generated by the RALS algorithm converges, then its limit point is a critical point of the original ALS minimization problem. This indicates that RALS can effectively solve some convergence problems in ALS. - **Numerical Experiments**: Through experiments on multiple data sets, it shows the advantages of RALS in reducing the swamp phenomenon and accelerating convergence, especially in the case of random initial factors. ### Formula Summary - **Sub - problems of ALS**: \[ A^{k + 1}=\arg\min_{\hat{A}\in\mathbb{R}^{I\times R}}\|T^{(1)}-\hat{A}(C^{k}\odot B^{k})^T\|_F^2 \] \[ B^{k + 1}=\arg\min_{\hat{B}\in\mathbb{R}^{J\times R}}\|T^{(2)}-\hat{B}(C^{k}\odot A^{k + 1})^T\|_F^2 \] \[ C^{k + 1}=\arg\min_{\hat{C}\in\mathbb{R}^{K\times R}}\|T^{(3)}-\hat{C}(B^{k + 1}\odot A^{k + 1})^T\|_F^2 \] - **Sub - problems of RALS**: \[ A^{k + 1}=\arg\min_{\hat{A}\in\mathbb{R}^{I\times R}}\|T^{(1)}-\hat{A}(C^{k}\odot B^{k})^T\|_F^2+\lambda_k\|A^{k}-\hat{A}\|_F^2 \] \[ B^{k + 1}=\arg\min_{\hat{B}\in\mathbb{R}^{J\times R}}\|T^{(2)}-\hat{B}(C^{k}\odot A^{k + 1})^T\|_F^2+\lambda_k\|B^{k}-\hat{B}\|_F^2 \] \[ C^{k + 1}=\arg\min_{\hat{C}\in\math

Some Convergence Results on the Regularized Alternating Least-Squares Method for Tensor Decomposition

Convergence of Projected Subgradient Method with Sparse or Low-Rank Constraints

On the Global Convergence of the Alternating Least Squares Method for Rank-One Approximation to Generic Tensors

Alternating minimization algorithms for graph regularized tensor completion

Randomized Alternating Least Squares for Canonical Tensor Decompositions: Application to a PDE with Random Data

Fast and Accurate Randomized Algorithms for Low-rank Tensor Decompositions

Efficient Alternating Least Squares Algorithms for Low Multilinear Rank Approximation of Tensors

Convergence Analysis of Graph Regularized Non-Negative Matrix Factorization

An alternating iterative minimisation algorithm for the double-regularised total least square functional

Convergence Analysis of a Variable Projection Method for Regularized Separable Nonlinear Inverse Problems

Accelerating regularized tensor decomposition using the alternating direction method of multipliers with multiple Nesterov's extrapolations

Stochastic Mirror Descent for Low-Rank Tensor Decomposition Under Non-Euclidean Losses

Escaping from Saddle Points --- Online Stochastic Gradient for Tensor Decomposition

Asymptotic Log-Det Sum-of-Ranks Minimization via Tensor (Alternating) Iteratively Reweighted Least Squares

Quantitative Convergence of Quadratically Regularized Linear Programs

Implicit Regularization and Entrywise Convergence of Riemannian Optimization for Low Tucker-Rank Tensor Completion

Convergence rates for critical point regularization

Low-Rank Tensor Completion via Novel Sparsity-Inducing Regularizers

Linearized Alternating Direction Method with Adaptive Penalty for Low-Rank Representation

R-Linear Convergence of Limited Memory Steepest Descent

Fast Learnings of Coupled Nonnegative Tensor Decomposition Using Optimal Gradient and Low-rank Approximation