Interpretable learning on knowledge bases

Experiments

Our aim was to generate interpretable rules from biological knowledge bases, mainly using algorithms comming from probabilistic logic programming.

We evaluate first order logic decision trees (TILDE) and probabilistic logic programs (Markov logic, Bayesian logic).

Implementations we tried include a Python implementation of TILDE (in this repo), LoMRF (a Scala implementation of Markov logic, https://anskarl.github.io/LoMRF/), ProbCog (Python implementation of Markov logic and Bayesian logic, https://github.com/opcode81/ProbCog).

Acknowledgments

To the DBCLS and BioHackathon 2019 sponsors and organizers to select and allow this project development.

References

https://www.ijcai.org/proceedings/2019/0843.pdf

https://homes.cs.washington.edu/~pedrod/papers/mlj05.pdf

https://cacm.acm.org/magazines/2019/7/237715-unifying-logical-and-statistical-ai-with-markov-logic/abstract