Research map / Graph learning / GNN architectures
Universal Prompt Tuning for Graph Neural Networks
Taoran Fang, Yunchao Mercer Zhang, Yang Yang, Chunping Wang, Lei CHEN
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 |
|---|---|---|---|
| Use GPF instead of fine-tuning | 5-layer GIN pretrained with Infomax · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 72.93 -> 74.36 | Improved |
| Use GPF-plus instead of fine-tuning | 5-layer GIN pretrained with AttrMasking · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 73.97 -> 75.76 | Improved |
| Use GPF instead of fine-tuning | 5-layer GIN pretrained with ContextPred · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 74.53 -> 75.86 | Improved |
| Use GPF instead of fine-tuning | 5-layer GIN pretrained with GCL · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 70.27 -> 70.28 | No clear effect |
| Use GPF-plus instead of fine-tuning | 5-layer GIN pretrained with GCL · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 70.27 -> 71.39 | Improved |
| Use GPF instead of fine-tuning | 5-layer GIN pretrained with EdgePred · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 72.72 -> 74.51 | Improved |
| Use GPF instead of GPPT | 5-layer GIN pretrained with EdgePred · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 61.80 -> 74.51 | Improved |
| Use GPF instead of GPPT (w/o ol) | 5-layer GIN pretrained with EdgePred · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 70.97 -> 74.51 | Improved |
| Use GPF instead of GraphPrompt | 5-layer GIN pretrained with EdgePred · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 61.20 -> 74.51 | Improved |
| Use GPF-plus instead of GPF | 5-layer GIN pretrained with Infomax · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 74.36 -> 74.79 | Improved |
| Use GPF-plus instead of GPF | 5-layer GIN pretrained with AttrMasking · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 75.76 -> 75.76 | No clear effect |
| Use GPF-plus instead of GPF | 5-layer GIN pretrained with ContextPred · Molecular and protein benchmarks (Avg.) | ROC-AUC (%): 75.86 -> 76.15 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GraphPrompt — GraphPrompt: Graph-Based Prompt Templates for Biomedical Synonym Prediction (AAAI 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Edge Prompt Tuning for Graph Neural Networks (ICLR 2025 · similarity 0.93)
- HeterGP: Bridging Heterogeneity in Graph Neural Networks with Multi-View Prompting (AAAI 2025 · similarity 0.91)
- GraphTOP: Graph Topology-Oriented Prompting for Graph Neural Networks (NeurIPS 2025 · similarity 0.90)
- One Prompt Fits All: Universal Graph Adaptation for Pretrained Models (NeurIPS 2025 · similarity 0.90)
- HG-Adapter: Improving Pre-Trained Heterogeneous Graph Neural Networks with Dual Adapters (ICLR 2025 · similarity 0.89)
- Spectral Reversal: Counteracting Singular Value Bias for Graph Prompting (NeurIPS 2026 · similarity 0.89)
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.
