Research map / Graph learning / Graph algorithms and clustering
Deep learning for graph matching
Learn graph matching with neural networks and combinatorial solvers.
42 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · an approach in Graph algorithms and clustering, keyword cluster “graph matching · matching graph · correspondence”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
13 papers in 2023–24 and 5 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 6.6% to 2.5%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2017 · A graph-theoretic approach to multitasking (NeurIPS 2017)
- 2018 · Deep Learning of Graph Matching (CVPR 2018 · 237 citations)
- 2019 · Graph Matching Networks for Learning the Similarity of Graph Structured Objects (ICML 2019 · 298 citations)
- 2020 · Learning Combinatorial Solver for Graph Matching (CVPR 2020 · 52 citations)
- 2021 · Learnable Graph Matching: Incorporating Graph Partitioning With Deep Feature Learning for Multiple Object Tracking (CVPR 2021 · 144 citations)
- 2022 · ClusterGNN: Cluster-Based Coarse-To-Fine Graph Neural Network for Efficient Feature Matching (CVPR 2022 · 97 citations)
- 2023 · Deep Learning of Partial Graph Matching via Differentiable Top-K (CVPR 2023 · 12 citations)
- 2024 · Discrete Cycle-Consistency Based Unsupervised Deep Graph Matching (AAAI 2024 · 2 citations)
- 2025 · Towards Unsupervised Training of Matching-based Graph Edit Distance Solver via Preference-aware GAN (NeurIPS 2025 · 0 citations)
- 2026 · Chaining 2-FWL GNNs for Combinatorial Graph Alignment (NeurIPS 2026 · 0 citations)
Earliest papers here
- A graph-theoretic approach to multitasking (NeurIPS 2017)
- Deep Learning of Graph Matching (CVPR 2018 · 237 citations)
- Graph Matching Networks for Learning the Similarity of Graph Structured Objects (ICML 2019 · 298 citations)
What has been tried
38 papers here report 389 measured changes (a component added, removed or swapped, compared with a baseline): helped 73%, no clear effect 7%, hurt 15%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| use a QAP solver instead of a LAP solver | PascalVOC Filtered · accuracy (%) | 59.7 -> 62.4 | Helped | Discrete Cycle-Consistency Based Unsupervised Deep Graph Matching (AAAI 2024 · 2 citations) |
| add softassign transition | Random matrices, matrix size 2000 · speedup ratio | 6.7x | Helped | Adaptive Softassign via Hadamard-Equipped Sinkhorn (CVPR 2024 · 2 citations) |
| add both filtering and calibration instead of filtering alone | HPatches · MMA @1px | 39.9 -> 52.3 | Helped | FC-GNN: Recovering Reliable and Accurate Correspondences from Interferences (CVPR 2024 · 2 citations) |
| Use expanding-window matching instead of Fixed1 matching | Cora · accuracy | 82.3 -> 82.9 | Helped | Navigating Complexity: Toward Lossless Graph Condensation via Expanding Window Matching (ICML 2024 · 1 citation) |
| replace eager multi-layer updates with lazy multi-round updates | AIDS, MUTAG, PTC-FM, PTC-FR, PTC-MM, PTC-MR · MAP | Node: 0.756 -> 0.825, 0.81 -> 0.851, 0.859 -> 0.888, 0.802 -> 0.855, 0.827 -> 0.838, 0.841 -> 0.874; Edge: 0.795 -> 0.84 | Helped | Iteratively Refined Early Interaction Alignment for Subgraph Matching based Graph Retrieval (NeurIPS 2024 · 1 citation) |
| Add exploration to current-best-solution recovery | AIDS700 · MAE | 0.549 -> 0.088 | Helped | Towards Unsupervised Training of Matching-based Graph Edit Distance Solver via Preference-aware GAN (NeurIPS 2025 · 0 citations) |
| Learn matching biases instead of fixing them equal | SPair-71K · matching F1-score | 56.5 ± 0.4 -> 57.7 ± 0.2 | Helped | Learning Partial Graph Matching via Optimal Partial Transport (ICLR 2025 · 0 citations) |
| replace M3C-hard's hard clustering with the relaxed indicator in M3C | Willow ObjectClass, 3 clusters × 8 images, 0 outlier · MA | 0.838 -> 0.884 | Helped | M3C: A Framework towards Convergent, Flexible, and Unsupervised Learning of Mixture Graph Matching and Clustering (ICLR 2024 · 0 citations) |
| Replace the VTG approximator with alternative GNN architectures | Graph configurations, (10×20) · competitive ratio | No clear effect | MAGNOLIA: Matching Algorithms via GNNs for Online Value-to-go Approximation (ICML 2024 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 11 papers of this approach from 2024–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Willow Object Class · 2 of 11 papers
- SPair-71K · 2 of 11 papers
- LINUX · 2 of 11 papers
- IMDB · 2 of 11 papers
- AIDS · 2 of 11 papers
- Mutag · 2 of 11 papers
Baselines
- NGMv2 · 3 of 11 papers
- GANN · 2 of 11 papers
- BBGM · 2 of 11 papers
- S-GWL · 2 of 11 papers
- SimGNN · 2 of 11 papers
- GraphSim · 2 of 11 papers
- GREED · 2 of 11 papers
- GEDGNN · 2 of 11 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 18 papers; from OpenAlex and the papers' affiliations).
- Shanghai Jiao Tong University · 5 papers
- Indian Institute of Technology Bombay · 2 papers
Use it
Download BibTeX of 32 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Optimization algorithms for graph matching (42 papers)
Elsewhere in the topic
- Neural combinatorial optimization on graphs (159 papers)
- Graph comparison and random features (155 papers)
- Graph embedding and spectral sparsification (126 papers)
- Continuous DAG structure learning (102 papers)
- Neural and local graph clustering (81 papers)
- Graph distance and retrieval (79 papers)
- Anchor-based multi-view clustering (31 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).