Combining several Hidden markov models

Question

I have 3 hidden markov models in python that I trained for each robot behaviours. Imagine I have 3 behaviours A, B and C. And for each of them I have a HMM model. Now I want to combine them together to have one HMM that can recognise a complex behaviour. A complex behaviour means combine A, B and C together. For example I give a sequence to the model and the results should be something like this. First B after that A after that C after .... Now I have a question. How should I combine 3 or several HMM together? Generally, is it possible or not? Thanks in advance for your answer

Answer 1

Having multiple HMM models together with independent states can be done with a Factorial Hidden Markov Model. The original paper is by Zoubin Ghahramani & Michael I. Jordan (1995)

Several implementations exists in Python, for example disaggregator.fhmm. It builds upon hmmlearn.