Dseptor

A Core-Guided Approach to Learning Optimal Causal Graphs

is developed by the Constraint Reasoning and Optimization Group at the Department of Computer Science, University of Helsinki, and implemented by Antti Hyttinen and Paul Saikko on top of the LMHS MaxSAT solver.

The main reference for dseptor is [1]. An online appendix to [1], including formal proofs and further details, is available here.

Usage

USAGE: ./dseptor <file> [options]

COMMAND LINE ARGUMENTS:

<file>	: Input filename in ... format.

COMMAND LINE OPTIONS:
-h        : Display this help message.
-v        : Display the version of the program.
-o <out>  : Output clauses to file <out> in wcnf format and exit.

Input format

The input is a text file, in which each line corresponds to an independence test 
result. Each line has the form:

i w x y c j

where

i     0 if the result is dependence or 1 if the result is independence
w     weight, a positive integer representing the reliability
x     the positive index of the first node (starting from 1) 
y     the positive index of the second node
c     integer presentation of the conditioning set bit vector
      0 = {}, 1 = {1}, 2= {2}, 3 = {1,2}, 4 = {3}, 5 = {1,3}, ...
j     integer presentation of the intervention set bit vector

Downloads

dseptor

The program will eventually be made available here.

Benchmarks

Benchmarks used in [1] are available here.

References

[1] A Core-Guided Approach to Learning Optimal Causal Graphs.
Antti Hyttinen, Paul Saikko, and Matti Järvisalo.
In Proceedings of the 26th International Joint Conference on Artificial Intelligence (IJCAI 2017), pages 645-651. AAAI Press, 2017.
[Publisher's version] [pdf] [abstract/bibtex]