SVM (Support Vector Machine) -- small datasets

For small datasets (e.g. 20 samples in total, divided into two groups of equal sizes), is it possible to carry out the SVM without training the classifier? If so, which Matlab commands would be used?

[Walter Roberson]

Please expand on what you mean by "carry out the SVM" ? With SVM, training is the name given to the process of finding the dividing hyperplane; before you know the hyperplane you do not have any way of using the hyperplane to predict classifications.

[Vi]

Thank you so much and sorry for my ignorance!...