I'm looking for some python implementation (in pure python or wrapping existing stuffs) of HMM and Baum-Welch.
Some ideas?
I've just searched in google and I've found really poor material with respect to other machine learning techniques. Why?
I'm looking for some python implementation (in pure python or wrapping existing stuffs) of HMM and Baum-Welch.
Some ideas?
I've just searched in google and I've found really poor material with respect to other machine learning techniques. Why?
The scikit-learn has an HMM implementation. It was until recently considered as unmaintained and its usage was discouraged. However it has improved in the development version. I cannot vouch for its quality, though, as I know nothing of HMMs.
Disclaimer: I am a scikit-learn developer.
Edit: we have moved the HMMs outside of scikit-learn, to https://github.com/hmmlearn/hmmlearn
You can find Python implementations on:
BTW: See Example of implementation of Baum-Welch on Stack Overflow - the answer turns out to be in Python.
Have you seen NLTK?
It has some classes that are suitable for this sort of thing, but somewhat application dependent.
http://www.nltk.org/api/nltk.tag.html#nltk.tag.hmm.HiddenMarkovModelTrainer
If you are looking for something more 'education oriented', I wrote toy trainer a while ago:
Some implementation of basic algorithms (including Baum-welch in python) are available here: http://ai.cs.umbc.edu/icgi2012/challenge/Pautomac/baseline.php
The General Hidden Markov Model library has python bindings and uses the Baum-Welch algorithm.