Assume you want to perform supervised learning and to predict the price of houses according to the size of houses, it is an example of a clustering problem. 

(NOTE: Deducted points for the wrong answer(s). 0 point for 'Not Answering').

For the learning rate in gradient descent, if it is too big , it will require a lot of training time (assuming there is a convergence).

(NOTE: Deducted points for the wrong answer(s). 0 point for 'Not Answering').

Like the name, the Logistic Regression algorithm mainly will be used to solve a regression problem.

(NOTE: Deducted points for the wrong answer(s). 0 point for 'Not Answering').

Choose methods for evaluating features for feature selection.

(NOTE: Multiple answers are allowed).

From the code snippet for a GridSearchCV based cross-validation for Ridge alpha

hyper-parameter, choose all the correct

answers.

Among the k-Fold cross validation (CV)  and leave-p-out CV techniques, which one belongs to the exhaustive (or more exhaustive) method.

(NOTE: Deducted points for the wrong answer(s). 0 point for 'Not Answering').

In CV (Cross-Validation) process, which error is more important for the choice of the best parameters (including hyper-parameters)?

(NOTE: Deducted points for the wrong answers. 0 point for 'Not Answering').

The information of the weight table (weight  and variance

estimates) can be visualized in a weight plot.

The following plot shows the results from a

Choose all correct interpretation/explanations.

From the code snippet for a GridSearchCV based cross-validation for multilayer perceptron (MLP) hyper-parameters, choose all the correct answers (assuming that the number of independent variables/features could be a large number (e.g., 100), and the number of instances are relatively small (e.g., 500)).

In CNN (Convolutional Neural Networks) model training phase, how the windows size and  the window stride in the Pooling layer can be built/selected?