Suppose that in the transformed space the feature vectors and corresponding labels are given as follows:

(−1, 0; +), (0, 1; +), (1, 0; +), (−1, 2; −), (−1, 3; −), (0, 3; −), (1, 5; −)

(a) Plot the feature vectors. Label the support vectors and solve for the margin.
What is the SVM decision rule in this case? How would it classify the feature vector (1,2)?

(b) Now we augment our data by adding a new point (−1,1;+). Repeat part (a).

(c) If there were another data point in the set (0,0;−), describe in broad terms how we would have to change our SVM algorithm. Why is this so?