freth251 / hmm Goto Github PK

Implementation of algorithms used in hidden markov models.

files:

• backward.py: an implementation of the backward algorithm that computes the likelihood of an observation given parameters. It is computed recursively backwards from time T to time 0.
• baum_welch.py: an implementation of the baum-welch algorithm used to estimate the parameters that maximize the likelihood of an observation.
• example.ipynb: examples on how to use the algorithms.
• forward.py: an implementation of the forward algorithm that computes the likelihood of an observation given parameters. It is computed recursively forwards from time 0 to time T.
• viterbi.py: an implementation of the viterbi algorithm that decodes the most likely hidden states that led to a sequence of observations.
• requirements.txt: contains required python modules to run the algorithms

To learn more about each algorithm check out my notes.