Prüfer: Prof. Nöth

Bayes Classifier

• What are posteriors/priors?
• Why it is optimal, how can we compute all probabilities?
• How can we estimate these parameters?
• Why it is not always a good idea to use the 0-1 cost function and therefore the missclassification?

Gaussian classifier

• Relationship to the Bayes classifier?
• How can it computed?
• How can we estimate the parameters?

EM-Algorithm

• Why and how does the EM converge?
• Compute the E- and M-step.

Exercise

• kNN example code. Describe what is going on.
• What are the sizes of the matrices?

Summary

Very kind and relaxed atmosphere. Focus on practical considerations! Describe the Matlab code step by step and focus on the Matlab syntax! Student friendly grading.