Low-rank Matrix Completion using Alternating Minimization

被引:0
|
作者
Jain, Prateek [1 ]
Netrapalli, Praneeth [1 ,2 ]
Sanghavi, Sujay [2 ]
机构
[1] Microsoft Res India, Bangalore, Karnataka, India
[2] Univ Texas Austin, Austin, TX 78712 USA
关键词
Matrix Completion Alternating Minimization;
D O I
暂无
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Alternating minimization represents a widely applicable and empirically successful approach for hiding low -rank matrices that best fit the given data. For example, for the problem of low-rank matrix completion, this method is believed to be one of the most accurate and efficient, and formed a major component of the winning entry in the Netflix Challenge [17]. In the alternating IllinilikatiOn approach, the low-rank target matrix is written in a 6i -linear form, i.e. X UV'; the algorithm then alternates between finding the best U and the best V. Typically, each alternating step in isolation is convex and tractable. However the overall problem becomes non convex and is prone to local minima. In fact, there has been almost no theoretical understanding of when this approach yields a good result. In this paper we present, one of the first theoretical analyses of the performance of alternating minimization for matrix completion, and the related problem of matrix sensing. For both these problems, celebrated recent results have shown that they become well-posed and tractable once certain (now standard) conditions are imposed on the problem. We show that alternating minimization also succeeds under similar conditions. Moreover, compared to existing results, our paper shows that alternating minimization guarantees faster (in particular, geometric) convergence to the true matrix, while allowing a significantly simpler analysis.
引用
收藏
页码:665 / 674
页数:10
相关论文
共 50 条
  • [1] Low-rank matrix completion using nuclear norm minimization and facial reduction
    Huang, Shimeng
    Wolkowicz, Henry
    JOURNAL OF GLOBAL OPTIMIZATION, 2018, 72 (01) : 5 - 26
  • [2] Low-rank matrix completion using nuclear norm minimization and facial reduction
    Shimeng Huang
    Henry Wolkowicz
    Journal of Global Optimization, 2018, 72 : 5 - 26
  • [3] Low-Rank Matrix Completion
    Chi, Yuejie
    IEEE SIGNAL PROCESSING MAGAZINE, 2018, 35 (05) : 178 - 181
  • [4] Structured alternating minimization for union of nested low-rank subspaces data completion
    Ashraphijuo M.
    Wang X.
    IEEE Journal on Selected Areas in Information Theory, 2020, 1 (03): : 632 - 644
  • [5] Reflection Removal Using Low-Rank Matrix Completion
    Han, Byeong-Ju
    Sim, Jae-Young
    30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017), 2017, : 3872 - 3880
  • [6] A Converse to Low-Rank Matrix Completion
    Pimentel-Alarcon, Daniel L.
    Nowak, Robert D.
    2016 IEEE INTERNATIONAL SYMPOSIUM ON INFORMATION THEORY, 2016, : 96 - 100
  • [7] DECENTRALIZED LOW-RANK MATRIX COMPLETION
    Ling, Qing
    Xu, Yangyang
    Yin, Wotao
    Wen, Zaiwen
    2012 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2012, : 2925 - 2928
  • [8] Adaptive Low-Rank Matrix Completion
    Tripathi, Ruchi
    Mohan, Boda
    Rajawat, Ketan
    IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2017, 65 (14) : 3603 - 3616
  • [9] Matrix Completion and Low-Rank SVD via Fast Alternating Least Squares
    Hastie, Trevor
    Mazumder, Rahul
    Lee, Jason D.
    Zadeh, Reza
    JOURNAL OF MACHINE LEARNING RESEARCH, 2015, 16 : 3367 - 3402
  • [10] Matrix completion and low-rank SVD via fast alternating least squares
    Hastie, Trevor
    Mazumder, Rahul
    Lee, Jason D.
    Zadeh, Reza
    Journal of Machine Learning Research, 2015, 16 : 3367 - 3402