Abstract:Exact recovery from contaminated visual data plays an important role in various tasks. By assuming the observed data matrix as the addition of a low-rank matrix and a sparse matrix, theoretic guarantee exists under mild conditions for exact data recovery. Practically matrix nuclear norm is adopted as a convex surrogate of the non-convex matrix rank function to encourage low-rank property and serves as the major component of recently-proposed Robust Principal Component Analysis (R-PCA). Recent endeavors have focused on enhancing the scalability of R-PCA to large-scale datasets, especially mitigating the computational burden of frequent large-scale Singular Value Decomposition (SVD) inherent with the nuclear norm optimization. In our proposed scheme, the nuclear norm of an auxiliary matrix is minimized instead, which is related to the original low-rank matrix by random projection. By design, the modified optimization entails SVD on matrices of much smaller scale, as compared to the original optimization problem. Theoretic analysis well justifies the proposed scheme, along with greatly reduced optimization complexity. Both qualitative and quantitative studies are provided on various computer vision benchmarks to validate its effectiveness, including facial shadow removal, surveillance background modeling and large-scale image tag transduction. It is also highlighted that the proposed solution can serve as a general principal to accelerate many other nuclear norm oriented problems in numerous tasks.

Nonmonotone variable projection algorithms for matrix decomposition with missing data

Variable Projection Algorithms: Theoretical Insights and A Novel Approach for Problems with Large Residual

A Nonconvex Approach to Low-Rank and Sparse Matrix Decomposition with Application to Video Surveillance.

An Alternating Direction Algorithm for Matrix Completion with Nonnegative Factors

Generalized Nonconvex Nonsmooth Low-Rank Matrix Recovery Framework with Feasible Algorithm Designs and Convergence Analysis.

Projective Reconstruction with Missing Data on Factorization Algorithm

Variable projection algorithms with sparse constraint for separable nonlinear models

A Unified Variance Reduction-Based Framework for Nonconvex Low-Rank Matrix Recovery

A Sequentially Truncated Higher Order Singular Value Decomposition-Based Algorithm for Tensor Completion

Projective Reconstruction with Missing Data Combining 2D Reprojection Error and Subspace Method

An algorithm for low-rank matrix factorization and its applications

The Application of ACA and SVD Algorithms in the Matrix Low-Rank Approximation

Joint Variable and Variable Projection Algorithms for Separable Nonlinear Models Using Aitken Acceleration Technique

Unified Solution to Nonnegative Data Factorization Problems

A Simple Decomposition Alternating Direction Method for Matrix Completion

Low-rank matrix recovery with non-quadratic loss: projected gradient method and regularity projection oracle

Accelerated Low-Rank Visual Recovery by Random Projection

Low-rank matrix decomposition in L1-norm by dynamic systems

Low-rank matrix estimation via nonconvex spectral regularized methods in errors-in-variables matrix regression

Successively Alternate Least Square for Low-Rank Matrix Factorization with Bounded Missing Data

Dynamic Spect Reconstruction from Few Projections: A Sparsity Enforced Matrix Factorization Approach