Research map / Graph learning / GNN architectures
Deep graph convolution and simplification
Make GCNs deeper and simpler with spectral analysis.
82 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in GNN architectures, keyword cluster “convolutional · gcns · spectral”
State History Benchmarks Tried Baselines Checklist Preprints Who Export
Where it stands
Few recent reports The tracked benchmarks have at most one report from 2025–26 that could beat the old record, too few to say whether they are still moving.
13 papers in 2023–24 and 6 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 2.7% to 1.2%.
Open questions the data points to
- Papers report GCN on Pubmed (accuracy) anywhere from 79 to 88.13: comparisons against it depend on whose number you take.
- Papers report MLP on Wisconsin (accuracy) anywhere from 64.2 to 85.49: comparisons against it depend on whose number you take.
No longer reported by papers here: Pubmed (accuracy, last in 2024); Citeseer (accuracy, last in 2024); Texas (accuracy (%), last in 2023); Chameleon (accuracy (%), last in 2023); Cornell (accuracy (%), last in 2023).
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
- 2018 · Graph Attention Networks (ICLR 2018 · 9,105 citations)
- 2019 · Simplifying Graph Convolutional Networks (ICML 2019 · 1,272 citations)
- 2020 · DropEdge: Towards Deep Graph Convolutional Networks on Node Classification (ICLR 2020 · 649 citations)
- 2021 · Beyond Low-frequency Information in Graph Convolutional Networks (AAAI 2021 · 533 citations)
- 2022 · Powerful Graph Convolutional Networks with Adaptive Propagation Mechanism for Homophily and Heterophily (AAAI 2022 · 111 citations)
- 2023 · Beyond Graph Convolutional Network: An Interpretable Regularizer-Centered Optimization Framework (AAAI 2023 · 19 citations)
- 2024 · GCNext: Towards the Unity of Graph Convolutions for Human Motion Prediction (AAAI 2024 · 28 citations)
- 2025 · FicGCN: Unveiling the Homomorphic Encryption Efficiency from Irregular Graph Convolutional Networks (ICML 2025 · 0 citations)
- 2026 · GI-GCN: Global Interacted Graph Convolutional Networks via Dominant Sets for Graph Classification (ICML 2026)
Earliest papers here
- Graph Attention Networks (ICLR 2018 · 9,105 citations)
- Adaptive Graph Convolutional Neural Networks (AAAI 2018 · 774 citations)
- FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling (ICLR 2018 · 451 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
Cora · accuracy (higher is better)
19 papers, 2018–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 83 | Graph Attention Networks (ICLR 2018 · 9,105 citations) |
| 2020 | 88.2 | DropEdge: Towards Deep Graph Convolutional Networks on Node Classification (ICLR 2020 · 649 citations) |
Pubmed · accuracy (higher is better)
19 papers, 2018–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 79 | Graph Attention Networks (ICLR 2018 · 9,105 citations) |
| 2020 | 91.7 | DropEdge: Towards Deep Graph Convolutional Networks on Node Classification (ICLR 2020 · 649 citations) |
Citeseer · accuracy (higher is better)
18 papers, 2018–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 72.5 | Graph Attention Networks (ICLR 2018 · 9,105 citations) |
| 2020 | 80.5 | DropEdge: Towards Deep Graph Convolutional Networks on Node Classification (ICLR 2020 · 649 citations) |
Texas · accuracy (%) (higher is better)
7 papers, 2020–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 67.57 | Geom-GCN: Geometric Graph Convolutional Networks (ICLR 2020 · 429 citations) |
| 2021 | 71.72 | Universal Graph Convolutional Networks (NeurIPS 2021 · 55 citations) |
| 2022 | 85.17 | Powerful Graph Convolutional Networks with Adaptive Propagation Mechanism for Homophily and Heterophily (AAAI 2022 · 111 citations) |
Chameleon · accuracy (%) (higher is better)
6 papers, 2020–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 60.9 | Geom-GCN: Geometric Graph Convolutional Networks (ICLR 2020 · 429 citations) |
| 2021 | 61.95 | Why Do Attributes Propagate in Graph Convolutional Neural Networks? (AAAI 2021 · 28 citations) |
| 2022 | 70.9 | Deformable Graph Convolutional Networks (AAAI 2022 · 25 citations) |
Cornell · accuracy (%) (higher is better)
6 papers, 2020–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 60.81 | Geom-GCN: Geometric Graph Convolutional Networks (ICLR 2020 · 429 citations) |
| 2021 | 76.72 | Why Do Attributes Propagate in Graph Convolutional Neural Networks? (AAAI 2021 · 28 citations) |
| 2022 | 85.95 | Deformable Graph Convolutional Networks (AAAI 2022 · 25 citations) |
What has been tried
73 papers here report 772 measured changes (a component added, removed or swapped, compared with a baseline): helped 66%, no clear effect 9%, hurt 20%, mixed 6%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Replace dynamic selection with one fixed graph-convolution type | Human3.6M, 1000ms · MPJPE (mm) | 108.7 vs 109.9–116.0 | Hurt | GCNext: Towards the Unity of Graph Convolutions for Human Motion Prediction (AAAI 2024 · 28 citations) |
| Vary the number of convolution layers from 2 to 3 or 4 | Cora and ADNI · accuracy | Hurt | Learning to Approximate Adaptive Kernel Convolution on Graphs (AAAI 2024 · 10 citations) | |
| add HyperDrop | Pubmed · accuracy | 79.4±0.1 -> 79.9±0.2 | Helped | Residual Hyperbolic Graph Convolution Networks (AAAI 2024 · 4 citations) |
| increase QGRN depth from 3 to 18 layers | AIDS · mean test accuracy | 99.50 -> 99.00 | Hurt | Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · 1 citation) |
| Add Aggregation Order Optimization (AOO) | Cora · latency | 47.65±3.76 -> 40.39±1.61 | Helped | FicGCN: Unveiling the Homomorphic Encryption Efficiency from Irregular Graph Convolutional Networks (ICML 2025 · 0 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| GCN | Pubmed | accuracy | 79 – 88.13 | 4 papers, e.g. CVPR 2021 and ICLR 2020 |
| MLP | Wisconsin | accuracy | 64.2 – 85.49 | 2 papers, e.g. NeurIPS 2021 and AAAI 2022 |
| GCN | Squirrel | accuracy (%) | 33.1 – 44.07 | 2 papers, e.g. NeurIPS 2021 and AAAI 2022 |
| JKNet | Chameleon | accuracy | 53.95 – 68.53 | 2 papers, e.g. NeurIPS 2021 and AAAI 2022 |
| MLP | Cornell | accuracy | 65.89 – 83.24 | 2 papers, e.g. NeurIPS 2021 and AAAI 2022 |
| GCN | Wisconsin | accuracy (%) | 47.73 – 51.4 | 2 papers, e.g. NeurIPS 2021 and NeurIPS 2023 |
What reviewers will expect
Datasets and baselines named in the 5 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Cora · 2 of 5 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 19 papers; from OpenAlex and the papers' affiliations).
- Huawei · 2 papers
Use it
Download BibTeX of 33 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
- Oversmoothing and heterophily (60 papers)
- Spectral GNN design (37 papers)
- Theory of graph convolutional networks (83 papers)
Elsewhere in the topic
- Graph structure learning (201 papers)
- GNN generalization theory (187 papers)
- Simple and scalable GNNs (131 papers)
- Expressive equivariant message passing (109 papers)
- Robustness, privacy, and fairness of GNNs (94 papers)
- Physics-inspired GNNs (81 papers)
- Graph rewiring and scalable message passing (72 papers)
- Weisfeiler-Leman expressivity hierarchies (66 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).