Feature Weighting Using Margin and Radius Based Error Bound Optimization in SVMs

The Support Vector Machine error bound is a function of the margin and radius. Standard SVM algorithms maximize the margin within a given feature space, therefore the radius is fixed and thus ignored