full
search.png

Search

close.png

Cancel

arrow
Volume 38, Issue 2
Robust Inexact Alternating Optimization for Matrix Completion with Outliers

Ji Li, Jian-Feng Cai & Hongkai Zhao

DOI: 10.4208/jcm.1809-m2018-0106

J. Comp. Math., 38 (2020), pp. 337-354.

Published online: 2020-02

[An open-access article; the PDF is free to any online user.]

Preview Full PDF 3620 37964

Cited by

google scholar semantic scholar
Export citation
  • Abstract

We investigate the problem of robust matrix completion with a fraction of observation corrupted by sparsity outlier noise. We propose an algorithmic framework based on the ADMM algorithm for a non-convex optimization, whose objective function consists of an $\ell_1$ norm data fidelity and a rank constraint. To reduce the computational cost per iteration, two inexact schemes are developed to replace the most time-consuming step in the generic ADMM algorithm. The resulting algorithms remarkably outperform the existing solvers for robust matrix completion with outlier noise. When the noise is severe and the underlying matrix is ill-conditioned, the proposed algorithms are faster and give more accurate solutions than state-of-the-art robust matrix completion approaches.

  • Keywords

Matrix completion, ADMM, Outlier noise, Inexact projection.

  • AMS Subject Headings

65K05, 90C06, 93C41

  • Copyright

COPYRIGHT: © Global Science Press

  • Email address

keelee@csrc.ac.cn (Ji Li)

jfcai@ust.hk (Jian-Feng Cai)

zhao@uci.edu (Hongkai Zhao)

  • BibTex
  • RIS
  • TXT
@Article{JCM-38-337, author = {Li , JiCai , Jian-Feng and Zhao , Hongkai}, title = {Robust Inexact Alternating Optimization for Matrix Completion with Outliers}, journal = {Journal of Computational Mathematics}, year = {2020}, volume = {38}, number = {2}, pages = {337--354}, abstract = {

We investigate the problem of robust matrix completion with a fraction of observation corrupted by sparsity outlier noise. We propose an algorithmic framework based on the ADMM algorithm for a non-convex optimization, whose objective function consists of an $\ell_1$ norm data fidelity and a rank constraint. To reduce the computational cost per iteration, two inexact schemes are developed to replace the most time-consuming step in the generic ADMM algorithm. The resulting algorithms remarkably outperform the existing solvers for robust matrix completion with outlier noise. When the noise is severe and the underlying matrix is ill-conditioned, the proposed algorithms are faster and give more accurate solutions than state-of-the-art robust matrix completion approaches.

}, issn = {1991-7139}, doi = {https://doi.org/10.4208/jcm.1809-m2018-0106}, url = {http://global-sci.org/intro/article_detail/jcm/14520.html} }
TY - JOUR T1 - Robust Inexact Alternating Optimization for Matrix Completion with Outliers AU - Li , Ji AU - Cai , Jian-Feng AU - Zhao , Hongkai JO - Journal of Computational Mathematics VL - 2 SP - 337 EP - 354 PY - 2020 DA - 2020/02 SN - 38 DO - http://doi.org/10.4208/jcm.1809-m2018-0106 UR - https://global-sci.org/intro/article_detail/jcm/14520.html KW - Matrix completion, ADMM, Outlier noise, Inexact projection. AB -

We investigate the problem of robust matrix completion with a fraction of observation corrupted by sparsity outlier noise. We propose an algorithmic framework based on the ADMM algorithm for a non-convex optimization, whose objective function consists of an $\ell_1$ norm data fidelity and a rank constraint. To reduce the computational cost per iteration, two inexact schemes are developed to replace the most time-consuming step in the generic ADMM algorithm. The resulting algorithms remarkably outperform the existing solvers for robust matrix completion with outlier noise. When the noise is severe and the underlying matrix is ill-conditioned, the proposed algorithms are faster and give more accurate solutions than state-of-the-art robust matrix completion approaches.

Li , JiCai , Jian-Feng and Zhao , Hongkai. (2020). Robust Inexact Alternating Optimization for Matrix Completion with Outliers. Journal of Computational Mathematics. 38 (2). 337-354. doi:10.4208/jcm.1809-m2018-0106
Copy to clipboard
BibteX RIS TXT
The citation has been copied to your clipboard