Abstract:Tensor Robust Principal Component Analysis (TRPCA) holds a crucial position in machine learning and computer vision. It aims to recover underlying low-rank structures and characterizing the sparse structures of noise. Current approaches often encounter difficulties in accurately capturing the low-rank properties of tensors and balancing the trade-off between low-rank and sparse components, especially in a mixed-noise scenario. To address these challenges, we introduce a Bayesian framework for TRPCA, which integrates a low-rank tensor nuclear norm prior and a generalized sparsity-inducing prior. By embedding the proposed priors within the Bayesian framework, our method can automatically determine the optimal tensor nuclear norm and achieve a balance between the nuclear norm and sparse components. Furthermore, our method can be efficiently extended to the weighted tensor nuclear norm model. Experiments conducted on synthetic and real-world datasets demonstrate the effectiveness and superiority of our method compared to state-of-the-art approaches.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is the difficulties encountered by existing tensor robust principal component analysis (TRPCA) methods in accurately capturing the low - rank characteristics of tensors and balancing the trade - off between low - rank and sparse components, especially in the mixed - noise scenario. Specifically, current methods have difficulty in effectively recovering the underlying low - rank structure in high - dimensional data and characterizing the sparse structure of noise. ### Specific description of the problem 1. **Capturing of low - rank characteristics**: Existing TRPCA methods have difficulty in accurately capturing the low - rank characteristics of tensors when dealing with high - dimensional data. This is because the rank of a tensor is not as clearly defined as that of a matrix. Various definitions of tensor rank (such as CP rank, Tucker rank, etc.) have their own advantages and disadvantages, but all have certain limitations. 2. **Balancing of low - rank and sparse components**: In practical applications, data often contains low - rank components and sparse noise components. How to find an appropriate balance point between the two is a challenge. Especially when the data is affected by mixed noise (for example, Gaussian noise and other types of noise), this problem becomes more complex. 3. **Difficulty in parameter selection**: In the TRPCA model, choosing appropriate regularization parameters (such as λ) is crucial for model performance. However, traditional methods (such as the discretization principle, L - curve method, generalized cross - validation, etc.) usually require iterative minimization and are customized for specific regularization forms, which makes them insufficient in dealing with complex tensor recovery problems. ### Solutions proposed in the paper To solve the above problems, the author introduced a TRPCA framework based on variational Bayesian inference (VBI). This framework improves existing methods in the following ways: 1. **Introduction of Bayesian framework**: By embedding the low - rank tensor nuclear norm prior and the generalized sparse - inducing prior into the Bayesian framework, this method can automatically determine the optimal tensor nuclear norm and achieve a balance between low - rank and sparse components. 2. **Flexible model extension**: This method can be efficiently extended to the weighted tensor nuclear norm model, thereby improving the flexibility and adaptability of the model. 3. **Efficient inference method**: By using the Laplace approximation and the MM framework, this method can effectively handle the computational complexity brought by non - Gaussian posterior distributions, thereby significantly improving the inference efficiency and accuracy of low - rank and sparse tensors and their hyperparameters. ### Summary The core objective of the paper is to improve the TRPCA method by introducing a variational Bayesian inference framework to better handle the low - rank and sparse structures in high - dimensional data and provide more robust solutions in the mixed - noise scenario. Experimental results show that this method outperforms existing methods on both synthetic data and real - world data sets.

Variational Bayesian Inference for Tensor Robust Principal Component Analysis

Robust Principal Component Analysis Based on Maximum Correntropy Criterion

Tensor Robust Principal Component Analysis Via Non-Convex Low Rank Approximation

Tensor Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Tensors via Convex Optimization

Tensor Robust Principal Component Analysis Via Non-convex Low-Rank Approximation Based on the Laplace Function

Tensor Robust Principal Component Analysis with a New Tensor Nuclear Norm

Convex–Concave Tensor Robust Principal Component Analysis

Low-Multi-Rank High-Order Bayesian Robust Tensor Factorization

The Nonconvex Tensor Robust Principal Component Analysis Approximation Model Via the Weighted ℓ P-Norm Regularization

An Efficient Variational Bayesian Inference Approach Via Studient's-t Priors for Acoustic Imaging in Colored Noises

A Hierarchical Variational Bayesian Approximation Approach in Acoustic Imaging

Global Weighted Tensor Nuclear Norm for Tensor Robust Principal Component Analysis

Robust Tensor Principal Component Analysis in All Modes

A Fast Correction Approach to Tensor Robust Principal Component Analysis

TIHN: Tensor Improved Huber Norm for low-rank tensor recovery

Tensor robust PCA with nonconvex and nonlocal regularization

Tensor Robust Kernel PCA for Multidimensional Data

Efficient Tensor Robust PCA under Hybrid Model of Tucker and Tensor Train

Improvement of robust tensor principal component analysis based on generalized nonconvex approach

Improved Robust Tensor Principal Component Analysis Via Low-Rank Core Matrix