WFOMC

Lifted knowledge compilation for Weighted First-Order Model Counting.

Download

Download most recent version (version info). This version supports Skolemization (for MLNs with existential quantifiers or lifted probabilistic logic programming), Weight Learning and Structure Learning.

Publications

W. Meert, G. Van den Broeck, and A. Darwiche. Lifted inference for probabilistic logic programs. In Proceedings of the 1st Workshop on Probabilsitic Logic Programming (PLP), 2014.
https://lirias.kuleuven.be/handle/123456789/460212
G. Van den Broeck, W. Meert, and A. Darwiche. Skolemization for Weighted First-Order Model Counting. In Proceedings of the 14th International Conference on Principles of Knowledge Representation and Reasoning (KR), 2014.
https://lirias.kuleuven.be/handle/123456789/444264
G. Van den Broeck, W. Meert, and A. Darwiche. Skolemization for Weighted First-Order Model Counting. arXiv preprint arXiv:1312.5378, 2013.
http://arxiv.org/pdf/1312.5378
G. Van den Broeck, W. Meert and J. Davis. Lifted Generative Parameter Learning. In Proceedings of the 3rd International workshop on Statistical Relational AI (StarAI), held at the 27th AAAI Conference, 2013.
G. Van den Broeck. Lifted Inference and Learning in Statistical Relational Models. PhD dissertation KU Leuven, 2013.
https://lirias.kuleuven.be/handle/123456789/373041
G. Van den Broeck and J. Davis. Conditioning in First-Order Knowledge Compilation and Lifted Probabilistic Inference. In Proceedings of the 26th AAAI Conference on Artificial Intelligence (AAAI), 2012.
https://lirias.kuleuven.be/handle/123456789/345667
G. Van den Broeck, A. Choi, A. Darwiche. Lifted relax, compensate and then recover: From approximate to exact lifted probabilistic inference. In Proceedings of the conference on Uncertainty in Artificial Intelligence (UAI), 2012
https://lirias.kuleuven.be/handle/123456789/351575
M. Jaeger, G. Van den Broeck. Liftability of probabilistic inference: Upper and lower bounds. In Proceedings of the 2nd International Workshop on Statistical Relational AI (StarAI), 2012. https://lirias.kuleuven.be/handle/123456789/352388
W. Meert, G. Van den Broeck, N. Taghipour, D. Fierens, H. Blockeel, J. Davis, L. De Raedt. Lifted inference for probabilistic programming. In Proceedings of the NIPS Probabilistic Programming Workshop, 2012.
https://lirias.kuleuven.be/handle/123456789/369419
G. Van den Broeck. On the completeness of first-order knowledge compilation for lifted probabilistic inference. In Proceedings of the Annual Conference on Neural Information Processing Systems (NIPS), 2011
https://lirias.kuleuven.be/handle/123456789/316338
G. Van den Broeck, N. Taghipour, W. Meert, J. Davis, and L. De Raedt. Tutorial on Lifted Inference in Probabilistic Logical Models. On the 22th International Joint Conference on Artificial Intelligence (IJCAI), 2011.
https://lirias.kuleuven.be/handle/123456789/317055
G. Van den Broeck, N. Taghipour, W. Meert, J. Davis, and L. De Raedt. Lifted probabilistic inference by first-order knowledge compilation. In Proceedings of the 22th International Joint Conference on Artificial Intelligence (IJCAI), 2011.
https://lirias.kuleuven.be/handle/123456789/308265

Example

A theory using the Markov Logic Network syntax can be parsed by WFOMC and queried as follows (> is the prompt).

# Example theory file
> cat models/friendsmoker.mln
person = {Guy, Nima, Wannes, Jesse, Luc}
Friends(person,person)
Smokes(person)
2 Friends(x,y) ^ Smokes(x) => Smokes(y)

# Run a query on the theory
> java -jar ./wfomc-1.0.jar -q "Smokes(Guy)" ./models/friendsmoker.mln
Reading file using MLN syntax.
Compilation took 815 ms
evidence nnf size = 18
evidence smooth nnf size = 24
query nnf size = 33
query smooth nnf size = 45

Inference took 15 ms
evidence logWmc = 68.63908810719217 = log(6.450259808376127E29)
query logWmc = 67.94594092663222 = log(3.225129904188057E29)

P(Some(Smokes(Guy))) = 0.49999999999999906

# Visualize the circuit in a pdf
> java -jar ./wfomc-1.0-SNAPSHOT.jar --pdf ./models/friendsmoker.mln
> open ./nnfs/theory.smooth.nnf.pdf

Here you can find the visualizations for the non-verbose circuit and the verbose circuit for the theory without the query.