Consider a binary linear classification where the data points are two dimensional, i.e., xR2 and the labels y{1,1}. The training set consists of the following points: x1=(1,0)T,x2=(1,0)T,x3=(1,d)T,y1=+1y2=1y3=+1 Assume that d>0 is some constant. For the perceptron algorithm please treat the points that fall exactly on the decision boundary as mistakes and do the update accordingly. Assume that the initial weight is the all-zero vector. (a) Sketch the points in the x1x2 plane and show the labels. Is the data-set linearly separable? (b) Run the perceptron algorithm in the following order (1,2,3),(1,2,3), until it converges. Show the output of the perceptron algorithm in each step, sketch the resulting decision boundary, and find the resulting margin (distance of the decision boundary to the nearest training point)