Differentiable Inductive Logic Programming

Python and TensorFlow implementation of the paper "Learning Explanatory Rules from Noisy Data." Evans Richard and Edward Grefenstette. Journal of Artificial Intelligence Research 61 (2018): 1-64.

Paper: https://arxiv.org/pdf/1711.04574.pdf

DeepMind blog: https://deepmind.com/blog/learning-explanatory-rules-noisy-data/

To run the code you create a folder. In our case the folder example/. It has three files.

• facts.ilp : contains the facts
• negative.ilp: contains the negative examples
• positive.ilp: contains the positive examples

You can run the code by:

python run.py example