Proposition of a classification system “β − LS − SVM” and its application to medical data sets

Author

Dammak, Fatma ; Baccour, Leila

Author_Institution

REGIM-Lab.: Res. Groups in Intell. Machines, Univ. of Sfax, Sfax, Tunisia

fYear

2014

fDate

11-14 Aug. 2014

Firstpage

101

Lastpage

105

Abstract

We apply two techniques of classification, Least Squares Support Vector Machines (LS-SVM) and Sequential Minimum Optimization SVM (SMO-SVM) to some diseases: cancer, hepatitis, heart, thyroid, and diabetes, described in Benchmark data sets. To compare between these techniques, some kernel functions are used which are polynomial, linear, sigmoidal, Gaussian and beta. Therefore the classifier β - LS - SMV is selected according to its best results.

Keywords

cancer; least squares approximations; medical computing; optimisation; pattern classification; support vector machines; β-LS-SVM classification system; Gaussian kernel functions; SMO-SVM; benchmark data sets; beta kernel functions; cancer; diabetes; diseases; heart; hepatitis; least squares support vector machines; linear kernel functions; medical data sets; polynomial kernel functions; sequential minimum optimization; sigmoidal kernel functions; thyroid; Cancer; Diabetes; Diseases; Iris recognition; Kernel; Support vector machines; Training; Classification; Kernel function; LS-SVM; SMO-SVM; Support Vector Machines (SVM); medical data sets;

fLanguage

English

Publisher

ieee

Conference_Titel

Soft Computing and Pattern Recognition (SoCPaR), 2014 6th International Conference of

Conference_Location

Tunis

Type

conf

DOI

10.1109/SOCPAR.2014.7007989

Filename

7007989