Maryia-M comments

Repositories
Issues
Comments

Results 3 comments of


                                            Maryia-M

Support of approximate inference?

I think you can use Gibbs Sampling. I use it to predict all node values given evidence in my big network (rejection_sample or likelihood_weighted_sample).

Support of approximate inference?

I don't suggest anything about it's speed, but exact inference simply don't work because of MemoryError if >= 15 node values are not given (I mean if (nodes_count - evidence_size)>=...

Support of approximate inference?

@linzhp I predicted using Gibbs Sampling (it's approximate method). Exact inference causes MemoryError, not Gibbs.