Support Vector Machines (2): Dual & soft-margin forms

Lagrangian optimization for the SVM objective; dual form of the SVM; soft-margin SVM formulation; hinge loss interpretation