| Limitations of mixture models |
|
|
|
|
|
|
|
|
|
|
|
| • | Mixture models
assume that the whole of each data vector |
|||||||||
| was generated by
exactly one of the models in the mixture. |
||||||||||
| – | This
makes is easy to compute the posterior distribution |
|||||||||
| over
models when given a data vector. |
||||||||||
| – | But it
cannot deal with data in which there are several |
|||||||||
| things
going on at once. |
||||||||||
| mixture of
10 models |
| mixture of 100 models |
