Support vector machines with two support vectors

Marcin Owczarczuk

No 35, Working Papers from Department of Applied Econometrics, Warsaw School of Economics

Abstract: In this article we present a new class of support vector machines for binary classification task. Our support vector machines are constructed using only two support vectors and have very low Vapnik-Chervonenkis dimension, so they generalize well. Geometrically, our approach is based on searching of a proper pair of observations from different classes of explained variable. Once this pair is found the discriminant hyperplane becomes orthogonal to the line connecting these observations. This method deals well with data sets with large number of features and small number of observations like gene expression data. We illustrate the performance of our classification method using gene expression data and show that it is superior to other classifiers especially to diagonal linear discriminant analysis and k-nearest neighbor which achieved the lowest error rate in the previous studies of tumor classification.

Keywords: support vector machines; Vapnik-Chervonenkis dimension; microarray experiment; tumor classification. (search for similar items in EconPapers)
Pages: 9 pages
Date: 2009-05-01
New Economics Papers: this item is included in nep-ecm
References: Add references at CitEc
Citations:

Downloads: (external link)
http://kolegia.sgh.waw.pl/pl/KAE/struktura/IE/stru ... Papers/aewp02-09.pdf (application/pdf)
Our link check indicates that this URL is bad, the error code is: 404 Not Found (http://kolegia.sgh.waw.pl/pl/KAE/struktura/IE/struktura/ZES/Documents/Working_Papers/aewp02-09.pdf [301 Moved Permanently]--> https://kolegia.sgh.waw.pl/pl/KAE/struktura/IE/struktura/ZES/Documents/Working_Papers/aewp02-09.pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:wse:wpaper:35

Access Statistics for this paper

More papers in Working Papers from Department of Applied Econometrics, Warsaw School of Economics Contact information at EDIRC.
Bibliographic data for series maintained by Marcin Owczarczuk ( this e-mail address is bad, please contact ).