The classical paradigm for data modeling invariably assumes that an input/output partitioning of the data is a priori given. For linear models, this paradigm leads to computational problems of solving approximately overdetermined systems of linear equations. Examples of most simple data fitting problems, however, suggest that the a priori fixed input/output partitioning of the data may be inadequate: (1) the fitting criteria often depend implicitly on the choice of the input and output variables, which may be arbitrary, and (2) the resulting computational problems are ill-conditioned in certain cases. An alternative paradigm for data modeling, sometimes refered to as the behavioral paradigm, does not assume a priori fixed input/output partitioning of the data. The corresponding computational problems involve approximation of a matrix constructed from the data by another matrix of lower rank. The chapter proceeds with review of applications in systems and control, signal processing, computer algebra, chemometrics, psychometrics, machine learning, and computer vision that lead to low rank approximation problems. Finally, generic methods for solving low rank approximation problems are outlined.
Great Common Divisor Kernel Principal Component Analysis Behavioral Paradigm Hankel Matrix Classical Paradigm
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
This is a preview of subscription content, log in to check access.
Alter O, Golub GH (2006) Singular value decomposition of genome-scale mRNA lengths distribution reveals asymmetry in RNA gel electrophoresis band broadening. Proc Natl Acad Sci 103:11828–11833
Bishop C (2006) Pattern recognition and machine learning. Springer, Berlin
Botting B (2004) Structured total least squares for approximate polynomial operations. Master’s thesis, School of Computer Science, University of Waterloo
Buckheit J, Donoho D (1995) Wavelab and reproducible research. In: Wavelets and statistics. Springer, Berlin/New York
Kiers H (2002) Setting up alternating least squares and iterative majorization algorithms for solving various matrix optimization problems. Comput Stat Data Anal 41:157–170
Kim H, Park H (2007) Sparse non-negative matrix factorizations via alternating non-negativity-constrained least squares for microarray data analysis. Bioinformatics 23:1495–1502