Exact MAG Learning

Tools for learning Maximal Ancestral Graphs.

Developed by the Constraint Reasoning and Optimization Group
at the Department of Computer Science, University of Helsinki.

Includes two programs for learning Maximal Ancestral Graphs:
The first one, scorer, calculates Gaussian BIC local scores from a given CSV file.
The second program, search, finds an optimal Maximal Ancestral Graph for the BIC local scores.
Details on the algorithms can be found in [1].


The source code of both C++ implementations can be downloaded here.


[1] Learning Maximal Ancestral Graphs with Exact Search
Kari Rantanen, Antti Hyttinen and Matti Järvisalo.
In Proceedings of the 37th Conference on Uncertainty in Artifical Intelligence (UAI 2021),
pages ???—???.