Imagine you train a function f on a training dataset, and you notice that the error is very low on the training set but is quite high on the testing set. For this question, recall that the generalization error is the expected cost, which for the squared error is E[(f(X)Y)2] and for classification is the expected 01 cost. Amongst the following statements, select all that apply. Note: If you select a statement that is true and one that is not true, then the incorrect choice cancels out the correct choice. (a) The function f likely has low generalization error (generalizes well). (b) The function f has likely overfit to the data. (c) The function f has likely underfit on the data. (d) The function f likely has high generalization error (generalizes poorly)