Consider a binary prediction problem (e.g., y=1 if "superbowl" / y=0 "no games") with two binary features x, (e.g., "eat nothing") and x, (e.g., "watch TV"). (a) Describe the decision rule to predict y=1 given a data point (x1, x2). (Hint: This rule described by probabilities universal regardless of discriminative or generative models). (b) Now you have collected 100 data points (e.g., 100 people) like the following. Evaluate the eight joint probabilities p(x1,x2,y) after Maximum Likelihood Estimation. - x1=0, x2=0, y=0, counts=12 - x1=0, x2=1, y=0, counts=8 - x1=1, x2=0, y=0, counts=18 - x1=1, x2=1, y=0, counts=12 - x1=0, x2=0, y=1, counts=5 - x1=0, x2=1, y=1, counts=25 - x1=1, x2=0, y=1, counts=5 - x1=1, x2=1, y=1, counts=15