Research map / Graph learning / GNN architectures
Meta-Aggregator: Learning To Aggregate for 1-Bit Graph Neural Networks
Yongcheng Jing, Yiding Yang, Xinchao Wang, Mingli Song, Dacheng Tao
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 GNA meta aggregator to 1-bit GAT (ablation) | vanilla 1-bit GAT · ZINC | test MAE: 0.670 -> 0.592 | Improved |
| add ANA meta aggregator to 1-bit GAT (ablation) | vanilla 1-bit GAT · ZINC | test MAE: 0.670 -> 0.566 | Improved |
| add GNA meta aggregator to 1-bit GCN (ablation) | vanilla 1-bit GCN · ZINC | test MAE: 0.669 -> 0.608 | Improved |
| add ANA meta aggregator to 1-bit GCN (ablation) | vanilla 1-bit GCN · ZINC | test MAE: 0.669 -> 0.607 | Improved |
| binarize GCN to 1-bit without meta aggregators | full-precision GCN · ZINC | test MAE: 0.407 -> 0.669 | Hurt |
| sum outputs of five standard aggregators in 1-bit GCN (ablation) | vanilla 1-bit GCN · ZINC | test MAE: 0.669 -> 0.599 | Improved |
| concatenate outputs of several aggregators in 1-bit GCN (ablation) | vanilla 1-bit GCN · ZINC | test MAE: 0.669 -> 0.614 | Improved |
| replace GAT aggregator with ANA in full-precision model | full-precision GAT · ZINC | test MAE: 0.476 -> 0.325 | Improved |
| replace GAT aggregator with GNA in full-precision model | full-precision GAT · ZINC | test MAE: 0.476 -> 0.337 | Improved |
| add GNA to 1-bit GAT (ablation) | vanilla 1-bit GAT · PPI | micro F1: 92.68 -> 97.52 | Improved |
| add ANA to 1-bit GAT (ablation) | vanilla 1-bit GAT · PPI | micro F1: 92.68 -> 97.71 | Improved |
| add GNA to 1-bit DGCNN (ablation) | vanilla 1-bit DGCNN · ModelNet40 | overall accuracy (%): 74.19 -> 78.36 | Improved |
Closest papers by meaning
- Principal Neighbourhood Aggregation for Graph Nets (NeurIPS 2020 · similarity 0.91)
- Integrating Co-Training with Edge Discrimination to Enhance Graph Neural Networks Under Heterophily (AAAI 2025 · similarity 0.88)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.88)
- Sequential Signal Mixing Aggregation for Message Passing Graph Neural Networks (NeurIPS 2024 · similarity 0.87)
- Graph Metanetworks for Processing Diverse Neural Architectures (ICLR 2024 · similarity 0.87)
- Making Classic GNNs Strong Baselines Across Varying Homophily: A Smoothness–Generalization Perspective (NeurIPS 2025 · similarity 0.87)
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.
