Question 7 /10 points/ SVM, Quadratic Approximation, Dual form, Lagrange multipliers Now consider the following "quadratic version" (primal form) of the soft SVM classification problem (you can assume that the intercept b is absorbed in w) 2 2 subject to yi(xi, w) 21- i for i-1,...,m Ei 0 for i = 1, ,m What is its dual form? Notes: In general, should remove vacuous constraints. As w, are primal variables, they should not appear in the dual form Be sure to specify whether the dual is a max or a min Question 7 /10 points/ SVM, Quadratic Approximation, Dual form, Lagrange multipliers Now consider the following "quadratic version" (primal form) of the soft SVM classification problem (you can assume that the intercept b is absorbed in w) 2 2 subject to yi(xi, w) 21- i for i-1,...,m Ei 0 for i = 1, ,m What is its dual form? Notes: In general, should remove vacuous constraints. As w, are primal variables, they should not appear in the dual form Be sure to specify whether the dual is a max or a min