TU Dortmund
RDT-B
Structure
2,000Graphs
430Nodes
498Edges
2Classes
Coverage
108Results
1Variants
25Papers
- Model-Driven Graph Contrastive Learning Model-Driven Graph Contrastive Learning
- ICML 2025: Learn Beneficial Noise as Graph Augmentation Learn Beneficial Noise as Graph Augmentation ICML 2025
- Multi-Scale Harmonic Encoding for Feature-Wise Graph Message Passing Multi-Scale Harmonic Encoding for Feature-Wise Graph Message Passing
- Motif-driven Subgraph Structure Learning for Graph Classification Motif-driven Subgraph Structure Learning for Graph Classification
- Dual-perspective Cross Contrastive Learning in Graph Transformers Dual-perspective Cross Contrastive Learning in Graph Transformers
- ICML 2024: Community-Invariant Graph Contrastive Learning Community-Invariant Graph Contrastive Learning ICML 2024
- WWW 2024: Graph Contrastive Learning with Cohesive Subgraph Awareness Graph Contrastive Learning with Cohesive Subgraph Awareness WWW 2024
- Graph Self-Contrast Representation Learning Graph Self-Contrast Representation Learning
- ENGAGE: Explanation Guided Data Augmentation for Graph Representation Learning ENGAGE: Explanation Guided Data Augmentation for Graph Representation Learning
- ICML 2023: Towards Better Graph Representation Learning with Parameterized Decomposition & Filtering Towards Better Graph Representation Learning with Parameterized Decomposition & Filtering ICML 2023
- Graph Contrastive Learning with Personalized Augmentation Graph Contrastive Learning with Personalized Augmentation
- ICLR 2022: Agent-based Graph Neural Networks Agent-based Graph Neural Networks ICLR 2022
- ICML 2022: Let Invariant Rationale Discovery Inspire Graph Contrastive Learning Let Invariant Rationale Discovery Inspire Graph Contrastive Learning ICML 2022
- ICML 2022: Omni-Granular Ego-Semantic Propagation for Self-Supervised Graph Representation Learning Omni-Granular Ego-Semantic Propagation for Self-Supervised Graph Representation Learning ICML 2022
- ICML 2022: Self-Supervised Representation Learning via Latent Graph Prediction Self-Supervised Representation Learning via Latent Graph Prediction ICML 2022
- WWW 2022: SimGRACE: A Simple Framework for Graph Contrastive Learning without Data Augmentation SimGRACE: A Simple Framework for Graph Contrastive Learning without Data Augmentation WWW 2022
- ICLR 2021: From Stars to Subgraphs: Uplifting Any GNN with Local Structure Awareness From Stars to Subgraphs: Uplifting Any GNN with Local Structure Awareness ICLR 2021
- NeurIPS 2021: Weisfeiler and Lehman Go Cellular: CW Networks Weisfeiler and Lehman Go Cellular: CW Networks NeurIPS 2021
- Prototypical Graph Contrastive Learning Prototypical Graph Contrastive Learning
- NeurIPS 2021: Adversarial Graph Augmentation to Improve Graph Contrastive Learning Adversarial Graph Augmentation to Improve Graph Contrastive Learning NeurIPS 2021
- ICML 2021: Graph Contrastive Learning Automated Graph Contrastive Learning Automated ICML 2021
- ICML 2021: Weisfeiler and Lehman Go Topological: Message Passing Simplicial Networks Weisfeiler and Lehman Go Topological: Message Passing Simplicial Networks ICML 2021
- Graph Self-Supervised Learning: A Survey Graph Self-Supervised Learning: A Survey
- ICML 2020: GraphNorm: A Principled Approach to Accelerating Graph Neural Network Training GraphNorm: A Principled Approach to Accelerating Graph Neural Network Training ICML 2020
- ICLR 2018: How Powerful are Graph Neural Networks? How Powerful are Graph Neural Networks? ICLR 2018
2-class graph classification Alias for REDDIT-BINARY in some papers. Uses the released node/edge labels or attributes; several social datasets are featureless. No canonical split; report folds and any added node features. Evaluated by Accuracy.
Dataset details
Task
Alias for REDDIT-BINARY in some papers. The primary catalog metric is Accuracy.
Data
Alias for REDDIT-BINARY in some papers. The cataloged artifact reports 2,000 graphs, 2 target classes or tasks.
Features
The TU text format may contain discrete node or edge labels and optional continuous attributes. Social-network datasets commonly have no intrinsic node attributes, so papers often add constant or degree features; that choice changes the input.
Splits and evaluation
The catalog records these protocols or variants: 10-fold CV. Evaluation uses Accuracy (higher is better). Exact masks or folds must come from the cited release.
Quirks and pitfalls
TU datasets have no canonical train/test split. Report cross-validation folds, seeds, feature construction, and whether duplicate or isomorphic graphs were removed; published results can otherwise differ for protocol rather than model quality.
Sources
SOTA Progression
new SOTA
matches SOTA at that time
other proposed result
other baseline result
Leaderboard
| # | Model | Accuracy ↑ | Paper |
|---|---|---|---|
| 1 | OEPG | 0.9630
±
0.0090 | |
| 2 | MSH-GNN | 0.9580
±
0.0090 | |
| 3 | GNN-AK+ | 0.9480
±
0.0080 | |
| 4 | AgentNet | 0.9420
±
0.0120 | |
| 5 | GraphNorm | 0.9350
±
0.0210 | |
| 6 | GSO | 0.9340
±
0.0130 | |
| 7 | ESAN | 0.9330
±
0.0130 | |
| 8 | GIN + InstanceNorm | 0.9320
±
0.0170 | |
| 9 | CRaWl | 0.9280
±
0.0220 | |
| 10 | Wit-TopoPool | 0.9280
±
0.0110 | |
| 11 | DGCL | 0.9270
±
0.0020 | |
| 12 | GIN-0 | 0.9240
±
0.0250 | |
| 13 | GIN | 0.9240 | |
| 14 | GIN + BatchNorm | 0.9240
±
0.0250 | |
| 15 | CIN | 0.9240
±
0.0210 | |
| 16 | Sum--MLP (GIN-0) | 0.9240
±
0.0250 | |
| 17 | AD-GCL-OPT | 0.9235
±
0.0042 | |
| 18 | CTAug-GraphCL | 0.9228
±
0.0033 | |
| 19 | GIN-ε | 0.9220
±
0.0230 | |
| 20 | SIN | 0.9220
±
0.0100 | |
| 21 | CTAug-JOAO | 0.9219
±
0.0024 | |
| 22 | AD-GCL-FIX | 0.9206
±
0.0042 | |
| 23 | GraphCL | 0.9169
±
0.0070 | |
| 24 | JOAO | 0.9166
±
0.0059 | |
| 25 | PGCL | 0.9150
±
0.0070 | |
| 26 | RGCL | 0.9138
±
0.0040 | |
| 27 | HTC | 0.9130 | |
| 28 | GraphSC | 0.9108
±
0.0056 | |
| 29 | DC-GCL | 0.9104
±
0.0045 | |
| 30 | CI-GCL | 0.9080
±
0.0050 | |
| 31 | DRGCL | 0.9080
±
0.0030 | |
| 32 | EG-SimCLR | 0.9070
±
0.0046 | |
| 33 | LaGraph | 0.9040
±
0.0080 | |
| 34 | RGCL | 0.9034
±
0.0058 | |
| 35 | GraphSC-MSE | 0.9026
±
0.0080 | |
| 36 | MGCL | 0.9025
±
0.0039 | |
| 37 | SEGA | 0.9021
±
0.0070 | |
| 38 | AD-GCL | 0.9007
±
0.0085 | |
| 39 | Sum--1-Layer | 0.9000
±
0.0270 | |
| 40 | GPA | 0.8999
±
0.0032 | |
| 41 | HG-SimCLR | 0.8992
±
0.0053 | |
| 42 | DropGIN | 0.8990
±
0.0170 | |
| 43 | GCC | 0.8980 | |
| 44 | HG-Simsiam | 0.8963
±
0.0055 | |
| 45 | GraphCL | 0.8953
±
0.0084 | |
| 46 | SimGRACE | 0.8951
±
0.0089 | |
| 47 | GraphCL(G) | 0.8950 | |
| 48 | AutoGCL | 0.8931
±
0.0148 | |
| 49 | EG-Simsiam | 0.8917
±
0.0043 | |
| 50 | JOAOv2 | 0.8879
±
0.0065 |
SOTA performance
set a new SOTA at the time
matched the SOTA at the time