1. (1 point) Draw the diagram for a multi-class mapper function with LDA, Perceptron and Logistic Regression classifiers, and write down their respective objective functions. Make sure to properly label your variables. 2. (1 point) List 2 advantages and 2 challenges of using a greedy iterative learning algorithm (such as gradient descent) over analytical solution (such as eigendecomposition or by matrix pseudoinverse). 3. (1 point) Consider the following dataset with 2 classes. Between LDA, Perceptron and Logistic Regression, which, if any, are able to separate the two classes with small to zero misclassification error? 1.0 0.5- * * 0.0 xxx -0.5 -1.0 -1.0 -0.5 0.0 0.5 1.0 x1 Co