Module dictionary
source code
Generic HMM model implementation, using NLTK's probability
handling.
This is similar to jazzparser.utils.nltk.ngram.NgramModel, but is
specialized to HMMs (bigram models) and stores probability distributions
as dictionaries instead of estimating them from counts. It may be trained
from counts in a corpus, but these are thrown away once the model is
estimated.
This type of model may be used in Baum-Welch re-estimation, since the
probabilities can be updated, since they're not estimated from counts.
Baum-Welch training for this model type (and its subclasses) can be found
in jazzparser.utils.nltk.ngram.baumwelch.
Author:
Mark Granroth-Wilding <mark.granroth-wilding@ed.ac.uk>
|
|
DictionaryHmmModel
Like an NgramModel, but (a) restricted to be an HMM (order 2 and no
backoff) and (b) uses dictionary distributions, rather than
distributions generated from counts.
|
|
|
__package__ = 'jazzparser.utils.nltk.ngram' |