English

Towards a General Recipe for Combinatorial Optimization with Multi-Filter GNNs

Machine Learning 2024-11-26 v2 Artificial Intelligence Discrete Mathematics

Abstract

Graph neural networks (GNNs) have achieved great success for a variety of tasks such as node classification, graph classification, and link prediction. However, the use of GNNs (and machine learning more generally) to solve combinatorial optimization (CO) problems is much less explored. Here, we introduce GCON, a novel GNN architecture that leverages a complex filter bank and localized attention mechanisms to solve CO problems on graphs. We show how our method differentiates itself from prior GNN-based CO solvers and how it can be effectively applied to the maximum cut, minimum dominating set, and maximum clique problems in a unsupervised learning setting. GCON is competitive across all tasks and consistently outperforms other specialized GNN-based approaches, and is on par with the powerful Gurobi solver on the max-cut problem. We provide an open-source implementation of our work at https://github.com/WenkelF/copt.

Keywords

Cite

@article{arxiv.2405.20543,
  title  = {Towards a General Recipe for Combinatorial Optimization with Multi-Filter GNNs},
  author = {Frederik Wenkel and Semih Cantürk and Stefan Horoi and Michael Perlmutter and Guy Wolf},
  journal= {arXiv preprint arXiv:2405.20543},
  year   = {2024}
}

Comments

In Proceedings of the Third Learning on Graphs Conference (LoG 2024, Oral); 20 pages, 2 figures

R2 v1 2026-06-28T16:47:58.577Z