# RES-PCA: A Scalable Approach to Recovering Low-Rank Matrices

@article{Peng2019RESPCAAS, title={RES-PCA: A Scalable Approach to Recovering Low-Rank Matrices}, author={Chong Peng and Chenglizhao Chen and Zhao Kang and Jianbo Li and Qiang Shawn Cheng}, journal={2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, year={2019}, pages={7309-7317} }

Robust principal component analysis (RPCA) has drawn significant attentions due to its powerful capability in recovering low-rank matrices as well as successful appplications in various real world problems. The current state-of-the-art algorithms usually need to solve singular value decomposition of large matrices, which generally has at least a quadratic or even cubic complexity. This drawback has limited the application of RPCA in solving real world problems. To combat this drawback, in this… Expand

#### Figures, Tables, and Topics from this paper

#### 22 Citations

Robust principal component analysis: A factorization-based approach with linear complexity

- Computer Science
- Inf. Sci.
- 2020

This paper proposes a novel RPCA model based on matrix tri-factorization, which only needs the computation of SVDs for very small matrices and reduces the complexity of RPCA to be linear and makes it fully scalable. Expand

Self-Paced Two-dimensional PCA

- Computer Science
- AAAI
- 2021

This work proposes a novel method called SP2DPCA, which progresses from ‘easy’ to ‘complex’ samples and looks for optimal projection matrix and filters out outliers iteratively, demonstrating the robustness nature of this method. Expand

Scaled Simplex Representation for Subspace Clustering

- Medicine, Computer Science
- IEEE Transactions on Cybernetics
- 2021

The proposed SSR-based SC (SSRSC) model is reformulated as a linear equality-constrained problem, which is solved efficiently under the alternating direction method of multipliers framework and outperforms the state-of-the-art SC methods on accuracy. Expand

Self-paced Principal Component Analysis

- Computer Science, Mathematics
- ArXiv
- 2021

This work proposes a novel method called Self-paced PCA (SPCA), which finds an optimal projection matrix and filters out outliers iteratively and can improve the state-of-the-art results considerably. Expand

Feature Selection Embedded Robust K-Means

- Computer Science
- IEEE Access
- 2020

From extensive experimental results, it is observed that superior clustering performance to the baseline methods is observed, which implies the effectiveness of the proposed method. Expand

A Unified Weight Learning and Low-Rank Regression Model for Robust Face Recognition

- Computer Science
- ArXiv
- 2020

A unified sparse weight learning and low-rank approximation regression model is applied to the robust face recognition in the presence of varying types and levels of corruptions, such as random pixel corruptions and block occlusions, or disguise. Expand

Bio-Inspired Structure Representation Based Cross-View Discriminative Subspace Learning via Simultaneous Local and Global Alignment

- Computer Science
- Complex.
- 2020

A novel cross-view discriminative feature subspace learning method inspired by layered visual perception from human, which utilizes a separable low-rank self-representation model to disentangle the class and view structure layers. Expand

Dynamic background subtraction with masked RPCA

- Computer Science
- Signal Image Video Process.
- 2021

This thesis proposes masked RPCA to process backgrounds containing moving textures and compared the proposed method with state-of-art, end-to-end non-deep learning methods to demonstrate its advantages. Expand

Generalized Locally-Linear Embedding: A Neural Network Implementation

- Computer Science
- 2020

This work introduces an extra fully-connected layer whose weight works as a reconstruction coefficient (i.e., relation among the samples) so that the latent representation can well preserve the neighborhood structure in locally-linear embedding. Expand

Discriminative Ridge Machine: A Classifier for High-Dimensional Data or Imbalanced Data

- Computer Science, Medicine
- IEEE Transactions on Neural Networks and Learning Systems
- 2021

In this article, we introduce a discriminative ridge regression approach to supervised classification. It estimates a representation model while accounting for discriminativeness between classes,… Expand

#### References

SHOWING 1-10 OF 30 REFERENCES

A Fast Factorization-Based Approach to Robust PCA

- Computer Science, Mathematics
- 2016 IEEE 16th International Conference on Data Mining (ICDM)
- 2016

A novel factorization-based model of RPCA, which has a complexity of O(kdn), where k is an upper bound of thetrue rank, and can be used as a light-weight, scalable tool for RPCA in the absence of the precise value of the true rank. Expand

Non-convex Robust PCA

- Computer Science, Mathematics
- NIPS
- 2014

A new provable method for robust PCA, where the task is to recover a low-rank matrix, which is corrupted with sparse perturbations, which represents one of the few instances of global convergence guarantees for non-convex methods. Expand

The Augmented Lagrange Multiplier Method for Exact Recovery of Corrupted Low-Rank Matrices

- Mathematics, Computer Science
- ArXiv
- 2009

The method of augmented Lagrange multipliers (ALM) is applied to solve the Robust PCA problem, namely recovering a low-rank matrix with an unknown fraction of its entries being arbitrarily corrupted, and it is proved the necessary and sufficient condition for the inexact ALM to converge globally. Expand

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

- Computer Science, Mathematics
- NIPS
- 2009

It is proved that most matrices A can be efficiently and exactly recovered from most error sign-and-support patterns by solving a simple convex program, for which it is given a fast and provably convergent algorithm. Expand

Background Recovery by Fixed-Rank Robust Principal Component Analysis

- Computer Science, Mathematics
- CAIP
- 2013

Comprehensive tests show that, by fixing the rank of the low-rank matrix to a known value, the fixed-rank algorithm produces more reliable and accurate results than existing low- rank RPCA algorithm. Expand

Robust PCA Via Nonconvex Rank Approximation

- Mathematics, Computer Science
- 2015 IEEE International Conference on Data Mining
- 2015

This work proposes a nonconvex rank approximation that outperforms current state-of-the-art algorithms in both accuracy and efficiency and develops an efficient augmented Lagrange multiplier based optimization algorithm. Expand

Stable Principal Component Pursuit

- Mathematics, Computer Science
- 2010 IEEE International Symposium on Information Theory
- 2010

This result shows that the proposed convex program recovers the low-rank matrix even though a positive fraction of its entries are arbitrarily corrupted, with an error bound proportional to the noise level, the first result that shows the classical Principal Component Analysis, optimal for small i.i.d. noise, can be made robust to gross sparse errors. Expand

Robust principal component analysis?

- Computer Science, Mathematics
- JACM
- 2011

It is proved that under some suitable assumptions, it is possible to recover both the low-rank and the sparse components exactly by solving a very convenient convex program called Principal Component Pursuit; among all feasible decompositions, this suggests the possibility of a principled approach to robust principal component analysis. Expand

Robust PCA via Outlier Pursuit

- Computer Science, Mathematics
- IEEE Transactions on Information Theory
- 2012

This work presents an efficient convex optimization-based algorithm that it calls outlier pursuit, which under some mild assumptions on the uncorrupted points recovers the exact optimal low-dimensional subspace and identifies the corrupted points. Expand

Subspace Clustering Using Log-determinant Rank Approximation

- Mathematics, Computer Science
- KDD
- 2015

The log-determinant low-rank optimization method is used to solve subspace clustering problem, for which the method of augmented Lagrangian multipliers is applied to optimize this non-convex rank approximation-based objective function and closes-form solutions for all subproblems of minimizing different variables alternatively. Expand