This is an implementation of Hidden Markov Models. It is split into parts to demonstrate solutions to the three classical problems solved by HMM's:

1. Finding the probability of an observation vector Y.
2. "Uncovering" the hidden state sequence, given Y.
3. Training the model so that it maximizes the probability P(Y).

The implementation is accompanied by a formal document (in Russian) which is, in fact, a coursework, written by the author during his study in the St. Petersburg State University, in 2016.

To run it, you'd need a jupyter notebook. Github has a built-in preview for the .ipynb files, though. The corresponding .py file is also included, but it may be out of sync with the source .ipynb file.

If you are interested in technical details or perhaps lonely and want to chat, drop a mail to [email protected]. All the best to you.