Research map / Graph learning / Knowledge and temporal graphs
Heterogeneous graph learning
Learn from and explain heterogeneous graphs with transfer and relational methods.
66 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · an approach in Knowledge and temporal graphs, keyword cluster “scene · image · visual”
State History Tried 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.
21 papers in 2023–24 and 15 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 8.1% to 5.3%.
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 · Learning Graphical State Transitions (ICLR 2017 · 77 citations)
- 2018 · GraphBit: Bitwise Interaction Mining via Deep Reinforcement Learning (CVPR 2018 · 38 citations)
- 2019 · Graph Neural Network for Music Score Data and Modeling Expressive Piano Performance (ICML 2019 · 19 citations)
- 2020 · Dynamic Multiscale Graph Neural Networks for 3D Skeleton Based Human Motion Prediction (CVPR 2020 · 368 citations)
- 2021 · Modeling the Momentum Spillover Effect for Stock Prediction via Attribute-Driven Graph Attention Networks (AAAI 2021 · 105 citations)
- 2022 · Sparse Structure Learning via Graph Neural Networks for Inductive Document Classification (AAAI 2022 · 53 citations)
- 2023 · Simple and Efficient Heterogeneous Graph Neural Network (AAAI 2023 · 224 citations)
- 2024 · PreRoutGNN for Timing Prediction with Order Preserving Partition: Global Circuit Pre-training, Local Delay Learning and Attentional Cell Modeling (AAAI 2024 · 21 citations)
- 2025 · BrainMAP: Learning Multiple Activation Pathways in Brain Networks (AAAI 2025 · 2 citations)
- 2026 · What Makes a Desired Graph for Relational Deep Learning? (ICML 2026 · 0 citations)
Earliest papers here
- Learning Graphical State Transitions (ICLR 2017 · 77 citations)
- GraphBit: Bitwise Interaction Mining via Deep Reinforcement Learning (CVPR 2018 · 38 citations)
- Graph Neural Network for Music Score Data and Modeling Expressive Piano Performance (ICML 2019 · 19 citations)
What has been tried
53 papers here report 532 measured changes (a component added, removed or swapped, compared with a baseline): helped 69%, no clear effect 8%, hurt 21%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add global pre-trained auto-encoder | 21 circuits, test average · slack R²uf | 0.720 -> 0.933 | Helped | PreRoutGNN for Timing Prediction with Order Preserving Partition: Global Circuit Pre-training, Local Delay Learning and Attentional Cell Modeling (AAAI 2024 · 21 citations) |
| Add marginal ranking loss | Node ranking (dataset not specified) · SPEARMAN | 0.680 -> 0.755 (+11.03%) | Helped | Deep Structural Knowledge Exploitation and Synergy for Estimating Node Importance Value on Heterogeneous Information Networks (AAAI 2024 · 15 citations) |
| add ALGA to ALTER | ABIDE · ACC | 71.0~(1.2) -> 77.0~(1.0) | Helped | Long-range Brain Graph Transformer (NeurIPS 2024 · 14 citations) |
| Use GSDA instead of a plain GCN for global edges | Unknown outdoor scene with SIFT · mAP5° | 63.08 -> 64.63 | Helped | MGNet: Learning Correspondences via Multiple Graphs (AAAI 2024 · 10 citations) |
| Add aggregation and adaptation masks | Delaware River Basin · RMSE | 1.833 -> 1.795 | Helped | Fair Graph Learning Using Constraint-Aware Priority Adjustment and Graph Masking in River Networks (AAAI 2024 · 7 citations) |
| Add global-information completion to Metis-partitioned subgraphs | Ogbn-mag · test accuracy | 35.07 -> 47.08 | Helped | An Efficient Subgraph-Inferring Framework for Large-Scale Heterogeneous Graphs (AAAI 2024 · 5 citations) |
| Add dummy tables | AB Rating · RMSE | 1.0291 -> 0.9847 | Helped | 4DBInfer: A 4D Benchmarking Toolbox for Graph-Centric Predictive Modeling on RDBs (NeurIPS 2024 · 3 citations) |
| add the adaptive structure sequentializer | HCP-Task, HCP-Gender, HCP-Age, HCP-FI, HCP-WM · accuracy on classification; MAE on regression | w/o LR: 94.56, 78.09, 48.13, 3.84, 10.96; BrainMAP: 94.74, 78.92, 48.44, 3.81, 10.75 | Helped | BrainMAP: Learning Multiple Activation Pathways in Brain Networks (AAAI 2025 · 2 citations) |
| use semantic-tree aggregation with subtree attention instead of weighted child-sum | IMDB · Micro-F1 | 69.70±0.24 -> 70.92±0.29 | Helped | Heterogeneous Graph Neural Network on Semantic Tree (AAAI 2025 · 2 citations) |
| add temporal attention (TAT) | Uplink throughput data (35min) · accuracy | 0.6125 +/- 0.0250 -> 0.7000 +/- 0.0250 | Helped | Fine-Grained Graph Representation Learning for Heterogeneous Mobile Networks with Attentive Fusion and Contrastive Learning (AAAI 2025 · 1 citation) |
| replace basic relationship pattern generation with meta-path sampling | GaoFen-2 · PSNR | 49.541 -> 48.212 | Hurt | HetSSNet: Spatial-Spectral Heterogeneous Graph Learning Network for Panchromatic and Multispectral Images Fusion (ICML 2025 · 1 citation) |
| Add bagging with different train/validation splits for early stopping | single-table tasks · prediction performance | Helped | CARTE: Pretraining and Transfer for Tabular Learning (ICML 2024 · 1 citation) |
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 36 papers; from OpenAlex and the papers' affiliations).
- Shanghai Jiao Tong University · 4 papers
- Zhejiang University · 3 papers
- Tsinghua University · 3 papers
- Peking University · 3 papers
- University of Utah · 2 papers
- Tianjin University · 2 papers
- Chinese Academy of Sciences · 2 papers
- University of Chinese Academy of Sciences · 2 papers
- Huawei · 2 papers
- The Chinese University of Hong Kong · 2 papers
Use it
Download BibTeX of 39 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
- Graph reasoning for visual recognition (96 papers)
- Graph-based dialog systems (94 papers)
- Social and behavior graphs (87 papers)
- Scene graph generation (70 papers)
Elsewhere in the topic
- Continuous-time dynamic graph learning (132 papers)
- Knowledge graph embeddings (110 papers)
- Temporal knowledge graph reasoning (84 papers)
- Temporal graph neural architectures (69 papers)
- Logical query answering and rule reasoning (68 papers)
- Spatio-temporal graph traffic forecasting (60 papers)
- Inductive knowledge graph completion (38 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).