SVM weight vector, support vectors and decision boundary understanding

91 Views Asked by At

I am trying to understand the relationship between the weight vector, the support vectors, and the decision boundary. Suppose that I have, the direction of the decision hyperplane [0.5 0.5 0.5 0.5], positive support vector [1 2 3 4] and negative support vector [0 0 0 0]. How are they related mathematically?

Can I derive the weight vector based on this information?

Thank you, I am new to SVM.

1

There are 1 best solutions below

0
On

if your SVM is without kernels, then the decision hyperplane IS the weight vector. You don't need the support vectors anymore

if using kernels, then the mapping is in a higher dimension and there's no simple relationship. See a similar question I asked some time ago, and in there you will also find a free book for SVMs that helped me a lot in understanding: reverse mapping of SVM kernel into original feature space