11 - Mixture Models

How long did you work on the last exercise sheets?

• I didn't do it.

• < 5hours

• 5-8hours

• >8hours

K-means is a special case of

• cross validation

• ICA

• mixture models

• PCA

Mixture models is K-means + ...

• density estimation

• pairwise clustering

• SOM

• PCA

K-means is not a special case of ...

• pairwise clustering

• SOM

• mixture models

• LLE

K-means is a special case of Expectation-Maximization (EM) as well (T/F).

• True

• False

• That component q creates a datapoint.

• That component q created a specific point x.

• That data point x was created by component q.

• That x follows a Gaussian.

• That component q creates a datapoint.

• The portion of points created by q.

• That data point x was created by component q.

• That x follows a Gaussian.

What is the constraint in the optimization of mixture models?

• There is no constraint.