search

jmschrei / pomegranate

Fast, flexible and easy to use probabilistic modelling in Python.
http://pomegranate.readthedocs.org/en/latest/
MIT License
3.29k stars 590 forks source link

[Question] Labeled Training with HMMs #1016

Closed AKuederle closed 1 year ago

AKuederle commented 1 year ago

Hi,

In version <1.0 it was possible to train HMMs using labeld/partially labeled sequences. In 1.0 this functionality seems to be removed. Is that something you are planning to add back at some point? Or if not, could you provide some tips, how the implement something like this in "user space"

Thanks in advance!

AKuederle commented 1 year ago

After further reading the new documentation, it seems like this can be accomplished, by passing prior to the fit method of the HMM. Is that correct?

AKuederle commented 1 year ago

I just found the full example about labeled training (missed it when first scanning the new docs...).

However, I am confused by the expected shape of the priors. But I will open a more specific issue about this.