Reproduce “Graph Matching via Optimal Transport” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ali Saad-Eldin, Benjamin D. Pedigo, Carey E. Priebe, Joshua T. Vogelstein
A new graph matching algorithm, GOAT, developed by Johns Hopkins University, integrates Lightspeed Optimal Transport into the FAQ framework, yielding substantially faster processing for large graphs and improved accuracy for inexact matching scenarios. It demonstrates GOAT's ability to recover true node correspondences in noisy, dynamic real-world networks where prior methods often fail.
Key insights
- Lightspeed Optimal Transport (LOT) can effectively replace the Linear Assignment Problem (LAP) sub-solver in iterative graph matching algorithms, offering substantial speed gains while maintaining accuracy.
- Using a "soft" doubly stochastic assignment at each iteration, as provided by LOT, is more robust to ties and ambiguity in node correspondences compared to enforcing a strict permutation.
- Leveraging optimal transport allows for more accurate and robust matching of inexact, noisy, and stochastic graphs, which are common in real-world datasets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.