Which of the following statement about linear discriminant function is NOT correct? O A. Least square can be used for learning the deicsion boundary for linear classification O B. Least square is sensitive to outlier samples. O C. Fisher linear discriminant is achieved by minimizing ratio of inter-class separation to intra-class variance. O D. Perceptron learning is achieved by minimize the number of misclassified examples. QUESTION 19 Which of the following statement on Generative model and Discriminative model is NOT true? O A. Generate model can generate synthetic example data, while Discriminative model can only be used for classification O B. Generative model have less number parameters to be trained than discriminative model OC. Generative model requires good model of class distribution, while Discriminative model requires good model of decision boundary O D. Generative model need to make assumption about class-conditional distribution, while Discriminative model doesn't need. QUESTION 20 Given a distribution p(x, y), it is always (in theory) possible to compute ply|X). O True O False