Fast and Robust Recursive Algorithms for Separable Nonnegative Matrix Factorization

被引:182
作者
Gillis, Nicolas [1 ]
Vavasis, Stephen A. [2 ]
机构
[1] Univ Mons, Dept Math & Operat Res, Fac Polytech, B-7000 Mons, Hainaut, Belgium
[2] Univ Waterloo, Dept Combinator & Optimizat, Waterloo, ON N2L 3G1, Canada
基金
加拿大自然科学与工程研究理事会;
关键词
Nonnegative matrix factorization; algorithms; separability; robustness; hyperspectral unmixing; linear mixing model; pure-pixel assumption; SPARSE; MODEL;
D O I
10.1109/TPAMI.2013.226
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this paper, we study the nonnegative matrix factorization problem under the separability assumption ( that is, there exists a cone spanned by a small subset of the columns of the input nonnegative data matrix containing all columns), which is equivalent to the hyperspectral unmixing problem under the linear mixing model and the pure-pixel assumption. We present a family of fast recursive algorithms and prove they are robust under any small perturbations of the input data matrix. This family generalizes several existing hyperspectral unmixing algorithms and hence provides for the first time a theoretical justification of their better practical performance.
引用
收藏
页码:698 / 714
页数:17
相关论文
共 30 条
[11]  
BOARDMAN JW, 1994, INT GEOSCI REMOTE SE, P2369, DOI 10.1109/IGARSS.1994.399740
[12]  
Businger P., 1965, NUMER MATH, V7, P269, DOI DOI 10.1007/BF01436084
[13]   A Simplex Volume Maximization Framework for Hyperspectral Endmember Extraction [J].
Chan, Tsung-Han ;
Ma, Wing-Kin ;
Ambikapathi, ArulMurugan ;
Chi, Chong-Yung .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2011, 49 (11) :4177-4193
[14]   A new growing method for simplex-based endmember extraction algorithm [J].
Chang, Chein-I ;
Wu, Chao-Cheng ;
Liu, Wei-min ;
Ouyang, Yen-Chieh .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2006, 44 (10) :2804-2819
[15]   Exponential Inapproximability of Selecting a Maximum Volume Sub-matrix [J].
Civril, Ali ;
Magdon-Ismail, Malik .
ALGORITHMICA, 2013, 65 (01) :159-176
[16]   On selecting a maximum volume sub-matrix of a matrix and related problems [J].
Civril, Ali ;
Magdon-Ismail, Malik .
THEORETICAL COMPUTER SCIENCE, 2009, 410 (47-49) :4801-4811
[17]   MINIMUM-VOLUME TRANSFORMS FOR REMOTELY-SENSED DATA [J].
CRAIG, MD .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 1994, 32 (03) :542-552
[18]   A Convex Model for Nonnegative Matrix Factorization and Dimensionality Reduction on Physical Space [J].
Esser, Ernie ;
Moeller, Michael ;
Osher, Stanley ;
Sapiro, Guillermo ;
Xin, Jack .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2012, 21 (07) :3239-3252
[19]   ROBUSTNESS ANALYSIS OF HOTTOPIXX, A LINEAR PROGRAMMING MODEL FOR FACTORING NONNEGATIVE MATRICES [J].
Gillis, Nicolas .
SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 2013, 34 (03) :1189-1212
[20]  
Gillis N, 2012, J MACH LEARN RES, V13, P3349