See also problog : https://dtai.cs.kuleuven.be/problog/

ProbLog2 is our second generation engine to reason with the ProbLog language. The current engine builds on logic programming, knowledge compilation, the distribution semantics and probabilistic, graphical models. It allows you to:

- Compute marginal probabilities of any number of ground atoms in the presence of evidence.

- Learn the parameters of the ProbLog program from partial interpretations.

-Sample from a ProbLog program.

-Solve decision theoretic problems

Problog rocks!

There's one thing I've always wanted to ask though - the first verison of Problog was implemented in Prolog and distributed as a library for YAP. How come Problog2 is implemented in Python? It's a probabilistic Prolog so I'd have thought Prolog would be the natural choice for it? Especially so coming from Ku-Leuven!

Looks very intuitive, and nice tutorial.

Nice project. Does ProbLog2 support continuous probability distributions?

Nice paper! This could be a great tool to help debugging differential privacy [0] of an algorithm. Certainly much simpler than CertiPriv [1].

[0] https://www.cis.upenn.edu/~aaroth/Papers/privacybook.pdf [1] http://certicrypt.gforge.inria.fr/certipriv/

I found out about this through a comment on my website, that includes a simple example of how to use it:

https://lee-phillips.org/chicks/#comments

Can anyone explain how this contrasts to Stan or other PP?

I think Stan uses sampling-based Monte Carlo, whereas PSI is based on static program analysis.

I'm a Stan user and I guess I'm interested to know if this would solve similar problems.