Are SVMs= distance based or discriminant clasifiers?
I mean the kernel is supposed to measure similarity, but the separating hyperplane is actually very much like a discriminant’s threshold, right?