| Universal Graph Transformer Self-Attention Networks | ✓ Link | 95.62% | | U2GNN (Unsupervised) | 2019-09-26 |
| Template based Graph Neural Network with Optimal Transport Distances | ✓ Link | 84.3% | | TFGW ADJ (L=2) | 2022-05-31 |
| Learning Universal Graph Neural Network Embeddings With Aid Of Transfer Learning | ✓ Link | 84.20% | | DUGNN | 2019-09-22 |
| When Work Matters: Transforming Classical Network Structures to Graph CNN | | 83.16% | | G_DenseNet | 2018-07-07 |
| Are Powerful Graph Neural Nets Necessary? A Dissection on Graph Classification | ✓ Link | 81.50% | | GFN | 2019-05-11 |
| Provably Powerful Graph Networks | ✓ Link | 81.38% | | PPGN | 2019-05-27 |
| Are Powerful Graph Neural Nets Necessary? A Dissection on Graph Classification | ✓ Link | 81.34% | | GFN-light | 2019-05-11 |
| Factorizable Graph Convolutional Networks | ✓ Link | 81.2% | 81.2% | FactorGCN | 2020-10-12 |
| Accurate Learning of Graph Representations with Graph Multiset Pooling | ✓ Link | 80.74% | | GMT | 2021-02-23 |
| Mutual Information Maximization in Graph Neural Networks | ✓ Link | 80.71% | | sGIN | 2019-05-21 |
| Fast Graph Representation Learning with PyTorch Geometric | ✓ Link | 80.6% | | GCN | 2019-03-06 |
| GraphMAE: Self-Supervised Masked Graph Autoencoders | ✓ Link | 80.32% | | Self-supervised GraphMAE | 2022-05-22 |
| How Powerful are Graph Neural Networks? | ✓ Link | 80.2% | | GIN-0 | 2018-10-01 |
| Wasserstein Embedding for Graph Learning | ✓ Link | 79.8% | | WEGL | 2020-06-16 |
| Maximum Entropy Weighted Independent Set Pooling for Graph Neural Networks | ✓ Link | 79.66% | | MEWISPool | 2021-07-03 |
| Capsule Graph Neural Network | ✓ Link | 79.62% | | CapsGNN | 2019-05-01 |
| Strengthening structural baselines for graph classification using Local Topological Profile | ✓ Link | 79.4 ± 2.5 | 79.4 ± 2.5 | Local Topological Profile (LTP) | 2023-05-01 |
| Hierarchical Representation Learning in Graph Neural Networks with Node Decimation Pooling | ✓ Link | 79.1% | | NDP | 2019-10-24 |
| Segmented Graph-Bert for Graph Instance Modeling | ✓ Link | 78.42% | | SEG-BERT | 2020-02-09 |
| Universal Graph Transformer Self-Attention Networks | ✓ Link | 77.84% | | U2GNN | 2019-09-26 |
| PANDA: Expanded Width-Aware Message Passing Beyond Rewiring | ✓ Link | 77.8% | | R-GIN + PANDA | 2024-06-06 |
| Graph U-Nets | ✓ Link | 77.56% | | Graph U-Nets | 2019-05-11 |
| Graph Representation Learning via Hard and Channel-Wise Attention Networks | ✓ Link | 77.48% | | hGANet | 2019-07-05 |
| Hierarchical Graph Representation Learning with Differentiable Pooling | ✓ Link | 75.48% | | GNN (DiffPool) | 2018-06-22 |
| PANDA: Expanded Width-Aware Message Passing Beyond Rewiring | ✓ Link | 75.11% | | GIN + PANDA | 2024-06-06 |
| A Fair Comparison of Graph Neural Networks for Graph Classification | ✓ Link | 73.9% | | GraphSAGE | 2019-12-20 |
| An End-to-End Deep Learning Architecture for Graph Classification | ✓ Link | 73.76% | | DGCNN | 2018-04-29 |
| Deep Graph Kernels | | 73.09% | | DGK | 2015-08-10 |
| DiffWire: Inductive Graph Rewiring via the Lovász Bound | ✓ Link | 72.24% | | DiffWire | 2022-06-15 |
| Graph Classification with 2D Convolutional Neural Networks | | 71.76% | | 2D CNN | 2017-07-29 |
| PANDA: Expanded Width-Aware Message Passing Beyond Rewiring | ✓ Link | 71.4% | | R-GCN + PANDA | 2024-06-06 |
| DiffWire: Inductive Graph Rewiring via the Lovász Bound | ✓ Link | 69.87% | | CT-Layer | 2022-06-15 |
| An End-to-End Deep Learning Architecture for Graph Classification | ✓ Link | 69.45% | | DGCNN (sum) | 2018-04-29 |
| PANDA: Expanded Width-Aware Message Passing Beyond Rewiring | ✓ Link | 68.4% | | GCN + PANDA | 2024-06-06 |
| DGCNN: Disordered Graph Convolutional Neural Network Based on the Gaussian Mixture Model | | 68.34% | | DGCNN | 2017-12-10 |
| Understanding Attention and Generalization in Graph Neural Networks | ✓ Link | 66.97% | | Weak-supervised ChebyNet | 2019-05-08 |
| DiffWire: Inductive Graph Rewiring via the Lovász Bound | ✓ Link | 65.89% | | GAP-Layer (Ncut) | 2022-06-15 |
| A Non-Negative Factorization approach to node pooling in Graph Convolutional Neural Networks | | 65.0% | | 1-NMFPool | 2019-09-07 |
| DiffWire: Inductive Graph Rewiring via the Lovász Bound | ✓ Link | 64.47% | | GAP-Layer (Rcut) | 2022-06-15 |