N. Ancona (1999)
On margin and support vector separability in Support Vector Machines for Regression
Istituto Elaborazione Segnali ed Immagini, Bari, Italy.
We show that when $\epsilon$ is close to 0, there exists a complete analogy between SVMR and SVMR, and the $\epsilon$-tube plays the same role as the margin between classes. For every $\epsilon$, the set of support vectors found by SVMR is linearly separable in the feature space and the optimal approximating hyperplane is a separator for this set.