ABSTRACT
In the chapter on Support Vector Machines, or SVM, the basic idea is presented: Find the optimal linear boundary between two classes. SVM is based on a number of mathematical procedures that are not here discussed in detail, but can be found in references. The foundation of the SVM is to find the widest possible boundary between the two classes and then to lead the border in the middle. The boundary-band edges are determined by a subset of training examples, that is, by using the support vectors. Another feature presented by SVM is that if a real boundary is nonlinear, the original vectors can be converted to another space with more dimensions where a linear (hyperplane) boundary can already be used. SVM principles are graphically displayed. Finally, the use of SVM using R package on real data is shown.
