Could you explain me why in jSVMLearner with non linear kernel, the vector w is calculated as linear combination of the training vectors, as in case of linear kernel? Why it is not simply given the function value which is sufficient to have the output?
thanks!
ilaria
