Describe in detail linear support vector machine method based on the primal formulation (10 points) given in the class and the one based on the dual formulation (10 points). Explain how the hyperplane is trained and then how it is used to make predictions. Describe also kernelized versions of the method - the exact one using exact values of kernels (10 points) and the one based on random feature maps (10 points). As for the linear method, explain in detail how training is conducted and how predictions are made. What are the advantages of the kernelized version (10 points)