Overview

This repository contains code for the paper:

Bryan Wilder, Bistra Dilkina, Milind Tambe. Melding the Data-Decisions Pipeline: Decision-Focused Learning for Combinatorial Optimization. AAAI Conference on Artificial Intelligence. 2019.

@inproceedings{wilder2019melding,
 author = {Wilder, Bryan},
 title = {Melding the Data-Decisions Pipeline: Decision-Focused Learning for Combinatorial Optimization},
 booktitle = {Proceedings of the 33rd AAAI Conference on Artificial Intelligence},
 year = {2019}
}

Included are differentiable solvers for LPs and submodular maximization, along with code to run the experiments in the paper. You can download the datasets from my website.

UPDATE: For the cora experiments, you can find the features here and the graph structures here.

Dependencies

• The linear programming experiments use the Gurobi solver.
• All code in the directory qpthlocal is derived from the qpth library. It has been modified to support use of the Gurobi solver in the forward pass.