Join GitHub today
GitHub is home to over 40 million developers working together to host and review code, manage projects, and build software together.Sign up
Implement GaussianMRF distribution #1981
This implements a
Note this Markov Random Field model involves an underlying factor graph, whereas the more standard Hidden Markov Model involves conditional normal distributions. This MRF implementation is simpler to implement than an HMM and is closer in code to the
The main parts to review are: new methods of the
@fehiepsi I'm pretty happy with this PR now. The last
In follow-up PRs I'd like to add an example, test jit support, test cuda support, etc.
fehiepsi left a comment
The implementation looks beautiful to me! I just catch a few nits during reviewing. Regarding to MRF log_prob, I think we can also test for identity transition_dist and observation_dist but it is not important IMO.