Problem 2. Boosting (30 points) Consider the dataset in the following figure. (a) What is the minimum number of iterations needed for AdaBoost with linear threshold functions as weak learners (e.g., sign(xi -c) or sign(c)) to achieve zero training error? (10 points) 2 (b) Apply Adaboost by hand on this example and write out the final classifier produced after 4 rounds of boosting. Show the weak classifiers and weights. (20 points) Problem 2. Boosting (30 points) Consider the dataset in the following figure. (a) What is the minimum number of iterations needed for AdaBoost with linear threshold functions as weak learners (e.g., sign(xi -c) or sign(c)) to achieve zero training error? (10 points) 2 (b) Apply Adaboost by hand on this example and write out the final classifier produced after 4 rounds of boosting. Show the weak classifiers and weights. (20 points)