Factor matrix trace norm minimization for low-rank tensor completion

Yuanyuan Liu, Fanhua Shang, Hong Cheng, James Cheng, Hanghang Tong

Research output: Chapter in Book/Report/Conference proceedingConference contribution

12 Citations (Scopus)

Abstract

Most existing low-n-rank minimization algorithms for tensor completion suffer from high computational cost due to involving multiple singular value decompositions (SVDs) at each iteration. To address this issue, we propose a novel factor matrix trace norm minimization method for tensor completion problems. Based on the CANDECOMP/PARAFAC (CP) decomposition, we first formulate a factor matrix rank minimization model by deducing the relation between the rank of each factor matrix and the mode-n rank of a tensor. Then, we introduce a tractable relaxation of our rank function, which leads to a convex combination problem of much smaller scale matrix nuclear norm minimization. Finally, we develop an efficient alternating direction method of multipliers (ADMM) scheme to solve the proposed problem. Experimental results on both synthetic and real-world data validate the effectiveness of our approach. Moreover, our method is significantly faster than the state-of-the-art approaches and scales well to handle large datasets.

Original languageEnglish (US)
Title of host publicationSIAM International Conference on Data Mining 2014, SDM 2014
PublisherSociety for Industrial and Applied Mathematics Publications
Pages866-874
Number of pages9
Volume2
ISBN (Print)9781510811515
DOIs
StatePublished - 2014
Externally publishedYes
Event14th SIAM International Conference on Data Mining, SDM 2014 - Philadelphia, United States
Duration: Apr 24 2014Apr 26 2014

Other

Other14th SIAM International Conference on Data Mining, SDM 2014
CountryUnited States
CityPhiladelphia
Period4/24/144/26/14

Fingerprint

Tensors
Singular value decomposition
Costs

ASJC Scopus subject areas

  • Computer Science Applications
  • Software

Cite this

Liu, Y., Shang, F., Cheng, H., Cheng, J., & Tong, H. (2014). Factor matrix trace norm minimization for low-rank tensor completion. In SIAM International Conference on Data Mining 2014, SDM 2014 (Vol. 2, pp. 866-874). Society for Industrial and Applied Mathematics Publications. https://doi.org/10.1137/1.9781611973440.99

Factor matrix trace norm minimization for low-rank tensor completion. / Liu, Yuanyuan; Shang, Fanhua; Cheng, Hong; Cheng, James; Tong, Hanghang.

SIAM International Conference on Data Mining 2014, SDM 2014. Vol. 2 Society for Industrial and Applied Mathematics Publications, 2014. p. 866-874.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Liu, Y, Shang, F, Cheng, H, Cheng, J & Tong, H 2014, Factor matrix trace norm minimization for low-rank tensor completion. in SIAM International Conference on Data Mining 2014, SDM 2014. vol. 2, Society for Industrial and Applied Mathematics Publications, pp. 866-874, 14th SIAM International Conference on Data Mining, SDM 2014, Philadelphia, United States, 4/24/14. https://doi.org/10.1137/1.9781611973440.99
Liu Y, Shang F, Cheng H, Cheng J, Tong H. Factor matrix trace norm minimization for low-rank tensor completion. In SIAM International Conference on Data Mining 2014, SDM 2014. Vol. 2. Society for Industrial and Applied Mathematics Publications. 2014. p. 866-874 https://doi.org/10.1137/1.9781611973440.99
Liu, Yuanyuan ; Shang, Fanhua ; Cheng, Hong ; Cheng, James ; Tong, Hanghang. / Factor matrix trace norm minimization for low-rank tensor completion. SIAM International Conference on Data Mining 2014, SDM 2014. Vol. 2 Society for Industrial and Applied Mathematics Publications, 2014. pp. 866-874
@inproceedings{4b54531e7ba9464bba8401a550af0829,
title = "Factor matrix trace norm minimization for low-rank tensor completion",
abstract = "Most existing low-n-rank minimization algorithms for tensor completion suffer from high computational cost due to involving multiple singular value decompositions (SVDs) at each iteration. To address this issue, we propose a novel factor matrix trace norm minimization method for tensor completion problems. Based on the CANDECOMP/PARAFAC (CP) decomposition, we first formulate a factor matrix rank minimization model by deducing the relation between the rank of each factor matrix and the mode-n rank of a tensor. Then, we introduce a tractable relaxation of our rank function, which leads to a convex combination problem of much smaller scale matrix nuclear norm minimization. Finally, we develop an efficient alternating direction method of multipliers (ADMM) scheme to solve the proposed problem. Experimental results on both synthetic and real-world data validate the effectiveness of our approach. Moreover, our method is significantly faster than the state-of-the-art approaches and scales well to handle large datasets.",
author = "Yuanyuan Liu and Fanhua Shang and Hong Cheng and James Cheng and Hanghang Tong",
year = "2014",
doi = "10.1137/1.9781611973440.99",
language = "English (US)",
isbn = "9781510811515",
volume = "2",
pages = "866--874",
booktitle = "SIAM International Conference on Data Mining 2014, SDM 2014",
publisher = "Society for Industrial and Applied Mathematics Publications",

}

TY - GEN

T1 - Factor matrix trace norm minimization for low-rank tensor completion

AU - Liu, Yuanyuan

AU - Shang, Fanhua

AU - Cheng, Hong

AU - Cheng, James

AU - Tong, Hanghang

PY - 2014

Y1 - 2014

N2 - Most existing low-n-rank minimization algorithms for tensor completion suffer from high computational cost due to involving multiple singular value decompositions (SVDs) at each iteration. To address this issue, we propose a novel factor matrix trace norm minimization method for tensor completion problems. Based on the CANDECOMP/PARAFAC (CP) decomposition, we first formulate a factor matrix rank minimization model by deducing the relation between the rank of each factor matrix and the mode-n rank of a tensor. Then, we introduce a tractable relaxation of our rank function, which leads to a convex combination problem of much smaller scale matrix nuclear norm minimization. Finally, we develop an efficient alternating direction method of multipliers (ADMM) scheme to solve the proposed problem. Experimental results on both synthetic and real-world data validate the effectiveness of our approach. Moreover, our method is significantly faster than the state-of-the-art approaches and scales well to handle large datasets.

AB - Most existing low-n-rank minimization algorithms for tensor completion suffer from high computational cost due to involving multiple singular value decompositions (SVDs) at each iteration. To address this issue, we propose a novel factor matrix trace norm minimization method for tensor completion problems. Based on the CANDECOMP/PARAFAC (CP) decomposition, we first formulate a factor matrix rank minimization model by deducing the relation between the rank of each factor matrix and the mode-n rank of a tensor. Then, we introduce a tractable relaxation of our rank function, which leads to a convex combination problem of much smaller scale matrix nuclear norm minimization. Finally, we develop an efficient alternating direction method of multipliers (ADMM) scheme to solve the proposed problem. Experimental results on both synthetic and real-world data validate the effectiveness of our approach. Moreover, our method is significantly faster than the state-of-the-art approaches and scales well to handle large datasets.

UR - http://www.scopus.com/inward/record.url?scp=84959917731&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84959917731&partnerID=8YFLogxK

U2 - 10.1137/1.9781611973440.99

DO - 10.1137/1.9781611973440.99

M3 - Conference contribution

SN - 9781510811515

VL - 2

SP - 866

EP - 874

BT - SIAM International Conference on Data Mining 2014, SDM 2014

PB - Society for Industrial and Applied Mathematics Publications

ER -