Question 1 (1 point) Perceptron learning rule can be used to update the weights of a machine learni model. True False Question 2 (1 point) In real applications, most data is linear-separable data. Therefore, we can use Al technique to analyze the data. True False Question 3 (1 point) SSE is a typical loss function that can be used to train a model. True False Question 4 (1 point) In gradient descent algorithm, the learning rate must be learned by machine itself. True False Question 5 (1 point) Logistic regression is typical method for regression problem. True False Question 6 (1 point) Which option is NOT a typical method for feature scaling or data standardization or normalization ? (difficult question) Assume the data is denoted by X. mu and std are the mean and standard deviation of X, respectively. The operation max() and min) compute the maximal and minimal value of X, respectively. (X - mu) / std X / max(X) -Jones: Attempt 1 Which option is NOT a typical method for feature scaling or data standardization or normalization ? (difficult question) Assume the data is denoted by X. mu and std are the mean and standard deviation of X, respectively. The operation max() and min() compute the maximal and minimal value of X, respectively. (X - mu) / std X / max(X) X* std (X - min(x))/(max(X) - min(X)) Question 7 (1 point) Which option is the logistic function used in logistic regression? 0(x) -X 1+ex Which statement is True about the Adaline and the logistic regression? O logistic regression adopts the sigmoid function as the activation function stochastic gradient descent cannot be used with logistic regression logistic regression solves regression problem by using gradient descent Adaline use non-linear activation function to improve performance Question 9 (1 point) Which option is NOT a hyper-parameter? number of training epochs learning rate model weight All of these options are hyper-parameters 3.0 2.5 2.0 Number of updates 1.5 1.0 0.5 0.0 2 8 10 Epochs at epoch 6 at epoch 2 at epoch 10 at epoch 4