TU Wien:Machine Learning VU (Musliu)/Exam 2025-06-25

Part 1 (26pt)

13 questions each when correct 2pt, when incorrect -1.pt, left empty 0pt

Part 2 (12pt)

4 questions each when correct 3pt, when incorrect -1.5pt, left empty 0pt

• From seven instances: Calculate/use Naive Bayesian to classify a new instance (binary)
• From seven instances: Calculate/use a Decision Tree with entroy to classify a new instance (binary)
• k bandit with sample averaging and Epsilon greedy: Given 8 periods with actions taken and gotten rewards, in which periods was a random action taken
• Given a regression function with w0, w1, w2 and 4 instances: Calculate the mean absoulte error
  • 2.5
  • 2
  • 1.5
  • None of the above

Part 3 (12pt)

• What are methods to deal with vanishing gradient?
• What has an effect on the size of the convolution layer in a CNN?
  • Increase padding, increases layer size
  • Decrease padding, increases layer size
  • Increase stride, increases layer size
  • Decrease stride, increases layer size
• Which of these are CNNs?
  • ResNet
  • TeNet
  • LeNet
  • LSMT
  • Rec... [forgot]
  • All of the above
  • None of the above