N. Ancona (1999)
Properties of Support Vector Machines for Regression
Istituto Elaborazione Segnali ed Immagini, Bari, Italy.
In this report we show that minimizing the norm of $w^2$ is equivalent to maximizing the sparsity of the representation of the optimal approximating hyperplane in SVMR. So the solution found by SVMR is a tradeoff between sparsity of the representation and closeness to the data.