Is er iets mis met deze pagina? Stuur dan naar


2017-2018: Jef Vandemeulebroucke

Exam questions

Example questions giving in last lecture

Question 1

Describe different types of CDSS based on:

a. The task they perform

b. The technology they are based on

c. The data they use as an input

Question 2

Describe Bayesian belief networks:

a. Explain the principle

b. What elements and data do you need to construct it?

c. How could you use it as a descriptive system?

d. Translate the following knowledge into a BBN (you’ll get a small text)

Question 3

Describe the k-NN classifier:

a. Describe the principle of the nearest neighbor rule

b. Explain the extension to k>1

c. Explain the process of training the classifier, and of applying the classifier.

d. How can the computation time be reduced?

e. How will error evolve on the training set and test set

  • for k=1
  • for k>1

Question 4

Describe the process of overfitting:

a. What is bias and variance?

b. What is statistical noise and deterministic noise?

c. Briefly describe tome methods to avoid overfitting

Question 5

Explain the difference between using machine learning CAD and deep learning CAD:

a. Which one requires the most data in your opinion? Why? (answer not black/white)

b. Which one will give you the most insight in image features being important? Why?

Question 6

Explain the principle of Bootstrapping:

a. Explain the principle

b. Give the steps when trying to estimate (a confidence interval of) the sensitivity

c. …


Bestand Commentaar