From Indefinite to Positive Semi-Definite Matrices

Similarity based classification methods use positive semi-definite (PSD) similarity matrices. When several data representations (or metrics) are available, they should be combined to build a single similarity matrix. Often the resulting combination is an indefinite matrix and can not be used to train the classifier. In this paper we introduce new methods to build a PSD matrix from an indefinite matrix. The obtained matrices are used as input kernels to train Support Vector Machines (SVMs) for classification tasks. Experimental results on artificial and real data sets are reported.

[1]  Catherine Blake,et al.  UCI Repository of machine learning databases , 1998 .

[2]  N. Higham.  Computing the nearest correlation matrix—a problem from finance , 2002 .

[3]  Robert P. W. Duin,et al.  A Generalized Kernel Approach to Dissimilarity-based Classification , 2002, J. Mach. Learn. Res..

[4]  John von Neumann,et al.  The geometry of orthogonal spaces , 1950 .

[5]  Javier M. Moguerza,et al.  Improving Support Vector Classification via the Combination of Multiple Sources of Information , 2004, SSPR/SPR.

[6]  Horst Bunke,et al.  On Not Making Dissimilarities Euclidean , 2004, SSPR/SPR.

[7]  U Emanuelson,et al.  A simple method for weighted bending of genetic (co)variance matrices. , 2003, Journal of dairy science.

[8]  Christopher J. Merz,et al.  UCI Repository of Machine Learning Databases , 1996 .

[9]  Erkki Oja,et al.  Artificial Neural Networks and Neural Information Processing — ICANN/ICONIP 2003 , 2003, Lecture Notes in Computer Science.

[10]  Javier M. Moguerza,et al.  Support Vector Machine Classifiers for Asymmetric Proximities , 2003, ICANN.

[11]  D. Luenberger.  Optimization by Vector Space Methods , 1968 .

[12]  W. G. Hill,et al.  Modification of Estimates of Parameters in the Construction of Genetic Selection Indices ('Bending') , 1981 .

[13]  Javier M. Moguerza,et al.  Combining Kernel Information for Support Vector Classification , 2004, Multiple Classifier Systems.

[14]  N. JARDINE,et al.  A New Approach to Pattern Recognition , 1971, Nature.