Anshul Dube
-
2009-05-19
Implementation of Hidden Markov Model with Gaussian Output
Brought to you by:
mokhov
This was my course project. Hope it is of some use.
This is implementation of Hidden Markov Model, the states emit output based on a multi-variate Gaussian distribution. The parameters required can be set in the main file. When the program is run it takes the training sequences from the training data and trains a HMM model for each class using Baum-Welch Algorithm, it calculates the log likelihood value based on the Viterbi Algorithm. Finally it generates a confusion matrix of classified examples.
It needs a HMM toolkit for operating, which can be found at: http://www.cs.ubc.ca/~murphyk/Software/HMM/hmm.html
This package is also open source.