Research map / Graph learning / Graph algorithms and clustering
Gromov-Wasserstein Learning for Graph Matching and Node Embedding
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Jointly learn node embeddings and graph matching instead of matching from observed graph distances alone | GWL-C vs GWD · MC3 Call→Email (Sparse) | Node Correctness (%): 23.16±0.46 -> 40.45±0.53 | Helped |
| Jointly learn node embeddings and graph matching instead of matching from observed graph distances alone | GWL-C vs GWD · MC3 Call→Email (Dense) | Node Correctness (%): 1.77±0.22 -> 4.23±0.27 | Helped |
| Use GWL-C rather than GWL-R embedding distances (ablation) | GWL · MC3 Call→Email (Sparse) | Node Correctness (%): 39.64±0.57 -> 40.45±0.53 | Helped |
| Use GWL-C rather than GWL-R embedding distances (ablation) | GWL · MC3 Call→Email (Dense) | Node Correctness (%): 3.80±0.23 -> 4.23±0.27 | Helped |
| Jointly learn matching and embeddings rather than use the one-outer-iteration GWD variant (ablation) | GWL-C vs GWD-C · MIMIC-III Top-1 | Precision (%): 43.16 -> 47.46 | Helped |
| Jointly learn matching and embeddings rather than use the one-outer-iteration GWD variant (ablation) | GWL-R vs GWD-R · MIMIC-III Top-1 | Precision (%): 46.29 -> 46.20 | No clear effect |
| Use cosine-based rather than RBF-based embedding distances (ablation) | GWL · MIMIC-III Top-1 | F1 (%): 22.22 -> 22.71 | Helped |
| Use cosine-based rather than RBF-based embedding distances (ablation) | GWL · MIMIC-III Top-5 | F1 (%): 34.18 -> 34.31 | Helped |
| Use GWL-C rather than Word2Vec | MIMIC-III · MIMIC-III Top-1 | F1 (%): 18.25 -> 22.71 | Helped |
| Use GWL-C rather than DWL (Finetune) | MIMIC-III · MIMIC-III Top-5 | Recall (%): 48.56 -> 45.64 | Hurt |
| Use cosine-based rather than RBF-based embedding distances (ablation) | GWD · MIMIC-III Top-1 | Precision (%): 46.29 -> 43.16 | Hurt |
Closest papers by meaning
- Scalable Gromov-Wasserstein Learning for Graph Partitioning and Matching (NeurIPS 2019 · similarity 0.93)
- Wasserstein Embedding for Graph Learning (ICLR 2021 · similarity 0.90)
- From One to All: Learning to Match Heterogeneous and Partially Overlapped Graphs (AAAI 2022 · similarity 0.90)
- Aligning Embeddings and Geometric Random Graphs: Informational Results and Computational Approaches for the Procrustes-Wasserstein Problem (NeurIPS 2024 · similarity 0.88)
- Learning Partial Graph Matching via Optimal Partial Transport (ICLR 2025 · similarity 0.88)
- THESAURUS: Contrastive Graph Clustering by Swapping Fused Gromov-Wasserstein Couplings (AAAI 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.