Post-Quiz on Linear Models

JavaScript isn't enabled in your browser, so this file can't be opened. Enable and reload.

Choose the best possible answers

Email *

Which of the following assumptions are made while arriving at the loss function for linear regression? *

1 point

The error terms between hypothesis and ground-truth are set to zero.

The error terms between hypothesis and ground-truth is assumed to be iid gaussian with zero mean and some variance.

The intercept terms are assumed to be iid gaussian with zero mean and some variance.

None of these

Suppose Y denotes the matrix of shape (N,1) which contains N ground-truth values in a linear regression problem. The matrix X is of shape (N,D) with each column denoting an independent feature for the same problem. Let N>D. The optimal solution for this setting, when calculated analytically, is given by? (Note: A^{-1} denotes the inverse of A whenever it exists and A' denotes the transpose of A when A is a matrix) *

1 point

(X' X)^{-1} X' Y

(XX')^{-1} X Y

(X'X)^{-1} Y (XX')

None

Which of the following is correct with respect to gradient descent algorithm? *

1 point

For each training example, you compute the gradient of loss function wrt each parameter and then update all the parameters at once

For each training example, you compute the gradient of loss function wrt each parameter , and then update all the parameters at once

For all the training examples, you compute the gradient of loss function wrt each parameter, and then update all the parameters at once.

For all the training examples, you compute the gradient of loss function wrt each parameter and update immediately.

The loss vs iterations curve for an optimization problem using gradient descent is follows. What are the possible reasons for this. *

1 point

Low learning rate

High learning rate

The learning rate is just right

None of the above

Which of the following regularization techniques has feature selecting property? *

1 point

Ridge

Lasso

Dropout

None of these

Which of the following linear classifiers is the least sensitive to outliers?

1 point

Logistic Regression

Perceptron

Linear SVM

Linear Regression

Clear selection

Submit

Clear form

This content is neither created nor endorsed by Google. Report Abuse - Terms of Service - Privacy Policy

Forms