The three initial problems of Hiden Markov Models
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Martins Eglitis 7440f1ec19 Update 2 years ago Initial commit 2 years ago Update 2 years ago


A script for solving the three initial problems of Hiden Markov Models

  • The evaluation problem (Forward algorithm, Backward algorithm) - what is the probability that the observations are generated by the model?
  • The decoding problem (Viterbi algorithm) - what is the most likely state sequence in the model that produced the observations?
  • The learning problem (Baum-Welch algorithm) - how should we adjust the model parameters A, B and Pi in order to maximize the probability?


A sample example is included basing on this article. Be aware that some models may produce errors in computations (e.g. division by zero) thus they need to be adjusted.