New Penalized Criteria for Smooth Non-Negative Tensor Factorization With Missing Entries

被引:0
作者
Durand, Amaury [1 ,2 ]
Roueff, Francois [1 ]
Jicquel, Jean-Marc [2 ]
Paul, Nicolas [3 ]
机构
[1] Inst Polytech Paris, LTCI, Telecom Paris, F-91120 Palaiseau, France
[2] Lab Renardieres, EDF R&D, TREE, E36, F-77818 Ecuelles, Moret Sur Loing, France
[3] PRISME, EDF R&D, F-78400 Chatou, France
关键词
Tensors; Signal processing algorithms; Optimization; Vectors; Convergence; Splines (mathematics); Research and development; Non-negative tensor decomposition; missing values; tensor completion; smoothness; PARAFAC; CP decomposition; MATRIX FACTORIZATION; PARAFAC; DECOMPOSITION; RANK;
D O I
10.1109/TSP.2024.3392357
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Tensor factorization models are widely used in many applied fields such as chemometrics, psychometrics, computer vision or communication networks. Real life data collection is often subject to errors, resulting in missing data. Here we focus in understanding how this issue should be dealt with for non-negative tensor factorization. We investigate several criteria used for non-negative tensor factorization in the case where some entries are missing. In particular we show how smoothness penalties can compensate the presence of missing values in order to ensure the existence of an optimum. This leads us to propose new criteria with efficient numerical optimization algorithms. Numerical experiments are conducted to support our claims.
引用
收藏
页码:2233 / 2243
页数:11
相关论文
共 35 条
[1]   Scalable tensor factorizations for incomplete data [J].
Acar, Evrim ;
Dunlavy, Daniel M. ;
Kolda, Tamara G. ;
Morup, Morten .
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS, 2011, 106 (01) :41-56
[2]   Structured regression models for high-dimensional spatial spectroscopy data [J].
Amini, Arash A. ;
Levina, Elizaveta ;
Shedden, Kerby A. .
ELECTRONIC JOURNAL OF STATISTICS, 2017, 11 (02) :4151-4178
[3]   The N-way Toolbox for MATLAB [J].
Andersson, CA ;
Bro, R .
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS, 2000, 52 (01) :1-4
[4]   PARAFAC. Tutorial and applications [J].
Bro, R .
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS, 1997, 38 (02) :149-171
[5]  
Cichocki R., 2009, Nonnegative Matrixand Tensor Factorizations: Applications to Exploratory Multi-Way DataAnalysis and Blind Source Separation
[6]  
Haeffele BD, 2015, Arxiv, DOI arXiv:1506.07540
[7]   TENSOR RANK AND THE ILL-POSEDNESS OF THE BEST LOW-RANK APPROXIMATION PROBLEM [J].
de Silva, Vin ;
Lim, Lek-Heng .
SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 2008, 30 (03) :1084-1127
[8]  
Durand A, 2021, EUR SIGNAL PR CONF, P1666, DOI 10.23919/EUSIPCO54536.2021.9615928
[9]   Are natural images of bounded variation? [J].
Gousseau, Y ;
Morel, JM .
SIAM JOURNAL ON MATHEMATICAL ANALYSIS, 2001, 33 (03) :634-648
[10]   Nonnegative Matrix Factorization over Continuous Signals using Parametrizable Functions [J].
Hautecoeur, Cecile ;
Glineur, Francois .
NEUROCOMPUTING, 2020, 416 :256-265