News for hackers. Stuff that matters.
Re:Pyro: PyTorch-Based Deep Universal Probabilistic Programming by nl on Friday November 03, 2017 @07:47PM (#15623021)
The combination of probalistic programming and deep learning is pretty interesting to me because that's what I have going on in two of my work projects.

What we do is have features built using deep learning models, then use that extract simple linear or categorical features which we condition our probalistic model on.

We've found it quite hard to use very high numbers of variables in the probalistic model.

Has anyone found a better way of doing this?

Re:Pyro: PyTorch-Based Deep Universal Probabilistic Programming by wakkaflokka on Friday November 03, 2017 @10:52PM (#15623544)
Do you have a recommended resource on how to feature engineer with DNNs for use as inputs in other models?
Re:Pyro: PyTorch-Based Deep Universal Probabilistic Programming by nl on Friday November 03, 2017 @11:02PM (#15623572)
I don't really understand the question.

For classification we just use the softmax over the classes. We had a long discussion about if this was close enough to probability to use and I think the conclusion was that it is.