search

ReactiveBayes / RxInfer.jl

Julia package for automated Bayesian inference on a factor graph with reactive message passing
MIT License
260 stars 24 forks source link

Question: What are some heuristic rules for initialization? #169

Closed svilupp closed 6 months ago

svilupp commented 2 years ago

It would be awesome to provide some simple tips for initmessages and initmarginals keywords in the inference().

It was briefly covered HERE, but it would be great to formalize the guidance a bit for people unfamiliar with the implementation.

albertpod commented 1 year ago

Should we transfer this issue to RxInfer.jl?

bvdmitri commented 1 year ago

Yes, the inference function is going to be transferred to the RxInfer.jl.

albertpod commented 8 months ago

What's the status of this one?

bvdmitri commented 8 months ago

@Nimrais & @mhidalgoaraya This task has been added to the milestone for tracking and prioritization.

Nimrais commented 8 months ago

What's the status of this one?

@albertpod, we had a discussion session with @mhidalgoaraya where we discussed several examples of situations where initialization is not necessary, and their counterparts where it is required. If I'm not mistaken, @mhidalgoaraya will prepare a Jupyter notebook or a markdown explanation.

albertpod commented 6 months ago

@mhidalgoaraya @Nimrais please update the ETA

mhidalgoaraya commented 6 months ago

@albertpod maybe I forgot to update but this was done several weeks ago, here it is the documentation https://reactivebayes.github.io/RxInfer.jl/stable/manuals/understanding-why-to-initialize-messages/