Research map / Graph learning / GNN architectures
BiKT: Unleashing the Potential of GNNs via Bi-Directional Knowledge Transfer
Shuai Zheng, Zhizhe Liu, Zhenfeng Zhu, Xingxing Zhang, Jianxin Li, Yao Zhao
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 BiKT bi-directional knowledge transfer | GCN · Cora (transductive) | accuracy: 73.44 -> 76.91 | Improved |
| add BiKT | GCN · Citeseer (transductive) | accuracy: 65.97 -> 68.63 | Improved |
| add BiKT | GCN · Pubmed (transductive) | accuracy: 81.62 -> 82.28 | Improved |
| add BiKT | MixHop · Citeseer (transductive) | accuracy: 62.64 -> 68.28 | Improved |
| add BiKT | GraphSAGE · OGB-Arxiv (transductive) | accuracy: 74.55 -> 75.47 | Improved |
| add BiKT | GraphSAGE · OGB-Products (inductive) | accuracy: 76.98 -> 77.47 | Improved |
| add BiKT | GCN · MUTAG | accuracy: 84.16 -> 87.26 | Improved |
| add BiKT | GIN · OGB-molhiv | accuracy: 76.09 -> 76.47 | No clear effect |
| train MLP with BiKT instead of plain MLP_GCN | MLP_GCN · Cora (transductive) | accuracy: 49.84 -> 74.99 | Improved |
| use BiKT-MLP_GCN instead of BiKT-GCN | BiKT-GCN · Citeseer (transductive) | accuracy: 68.63 -> 69.61 | Improved |
| add BiKT on disassortative nodes | GCN · Citeseer disassortative nodes | accuracy: 52.33 -> 65.14 | Improved |
| combine knowledge transfer and knowledge distillation terms (ablation) | BiKT · Cora and Citeseer | accuracy | Improved |
Closest papers by meaning
- Demystifying GNN-to-MLP Knowledge Transfer: Theoretical Grounding and Dual-Stream Distillation Method (AAAI 2026 · similarity 0.85)
- Can Classic GNNs Be Strong Baselines for Graph-level Tasks? Simple Architectures Meet Excellence (ICML 2025 · similarity 0.85)
- Fantastic Gains and Where to Find Them: On the Existence and Prospect of General Knowledge Transfer between Any Pretrained Model (ICLR 2024 · similarity 0.84)
- GraphTrail: Translating GNN Predictions into Human-Interpretable Logical Rules (NeurIPS 2024 · similarity 0.84)
- Classic GNNs are Strong Baselines: Reassessing GNNs for Node Classification (NeurIPS 2024 · similarity 0.84)
- GraphBridge: Towards Arbitrary Transfer Learning in GNNs (ICLR 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.
