I just learned the basics of SVM, and I have 2 questions about the solution for the non-separable case:

  1. In the separable case, the distance of all the support vectors from the hyperplane is the same, and they determine the margin. Now, in the non-separable case, is it possible that no support vector falls on the edge of the margin?

  2. In the separable case the meaning of $\frac{1}{\vert \vert \beta \vert \vert}$ was the distance of the support vectors from the hyperplane (where the hyperplane is defined by $\beta ^T x+\beta _0 = 0$). What is the exact meaning of $\frac{1}{\vert \vert \beta \vert \vert}$ in the non-separable case?

  • 1
    $\begingroup$ I would suggest you to look at literature that explaining SVN from hinge loss persopective $\endgroup$
    – Haitao Du
    Jan 16, 2017 at 17:26


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.