Quiz

the quiz every Monday covers the lecture content from the previous week.

Quiz 1

2/5

• (Multiple Choice): Which subset of the data is typically used to select hyperparameters (e.g., regularization strength, learning rate)?

  • A. Training set
  • B. Validation set
  • C. Test set
  • D. All of the above

• (True/False): Linear regression with L2 regularization has a closed form solution.

  Answer: True

• (True/False): When solving linear regression, gradient descent converges to a solution with lower loss than the closed-form solution.

  Answer: False

• (Short Answer): An L2-regularized linear regression model achieves low training error but significantly higher validation error. What does this indicate about the model’s bias and variance, and would you increase or decrease the regularization strength λ to address it?

  Answer: Low bias and high variance; increase the regularization strength (λ)

Quiz 2

5/5

1. True
2. Logit
3. False
4. A
5. 75

Quiz 3

3/3

Quiz 3 will cover only the two lectures on neural networks. Our schedule was pushed back by one lecture due to snow... KNN (along with decision tree) will be covered in the next Quiz.

Hi all, the grades for Quiz 3 published. Here are the correct answers:

• False
• B
• B