Research map / Graph learning / Graph self-supervised learning
HINPool: A Unified Heterogeneous Graph Pooling Framework for Accurate Molecular and Protein Property Prediction
Ming-Yi Hong, You-Chen Teng, Shao-En Lin, Chih-Yu Wang, Che Lin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add cross-layer fusion (ablation) | HINPool · MUTAG | AUROC: 60.8 -> 100.0 | Improved |
| add cross-layer fusion (ablation) | HINPool · ENZYMES | AUROC: 71.5 -> 92.8 | Improved |
| add type-aware selector (TAS) (ablation) | HINPool · PROTEINS | AUROC: 86.9 -> 89.8 | Improved |
| use type-aware Top-K instead of mixed Top-K (ablation) | HINPool · ENZYMES | AUROC: 92.7 -> 92.8 | No clear effect |
| add type-aware readout fusion (ablation) | HINPool · ENZYMES | AUROC: 73.2 -> 92.8 | Improved |
| model data as HIN with HINPool vs best baseline | HINPool · NCI109 | AUROC: 86.5 -> 86.2 | Hurt |
| model data as HIN with HINPool vs best baseline | HINPool · ENZYMES | AUROC: 86.2 -> 92.8 | Improved |
| reduce pooling ratio from 0.9 to 0.6-0.8 (ablation) | HINPool · PROTEINS | AUROC: 89.8 -> 88.3 (0.8), 88.6 (0.7), 88.5 (0.6) | Hurt |
| use 3 layers instead of 2 or 4 (ablation) | HINPool · PROTEINS | AUROC: 87.9 / 87.7 -> 89.8 | Improved |
| use declining pooling ratios 0.9/0.8/0.7 (ablation) | HINPool · PROTEINS | AUROC: 89.8 -> 88.4 | Hurt |
Closest papers by meaning
- Robust Heterogeneous Graph Classification for Molecular Property Prediction with Information Bottleneck (AAAI 2025 · similarity 0.89)
- Grouping Matrix Based Graph Pooling with Adaptive Number of Clusters (AAAI 2023 · similarity 0.86)
- InfoGraph: Unsupervised and Semi-supervised Graph-Level Representation Learning via Mutual Information Maximization (ICLR 2020 · similarity 0.86)
- Graph Parsing Networks (ICLR 2024 · similarity 0.84)
- Beyond Graphs: Can Large Language Models Comprehend Hypergraphs? (ICLR 2025 · similarity 0.84)
- Position: Graph Learning Will Lose Relevance Due To Poor Benchmarks (ICML 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
