Removing a Non-Support Vector (3 points each) Suppose we apply a binary SVM classifier with kernel k to a dataset (x1, y1), . . . ,(xn, yn). Suppose is a solution to the dual optimization problem, and that xi is not a support vector for a certain i. This implies i = 0. Now consider the modified data set obtained by removing (xi , yi). Let i be the length n 1 vector obtained by removing the i-th entry from . (a) Show that i solves the dual for the modified dataset. (b) Use part (a) to show that the SVM remains unchanged