2. (12 points) Select all the correct statement(s). (a) Non-parametric methods do not involve any parameters. (b) Parametric methods are always easier to compute in the test phase. (c) In k-Nearest Neighbor estimation, increasing k generally makes the prediction speed slower. (d) A linear discriminant function does not produce a linear decision boundary between every pair of classes in a classification task with more than 2 classes.