Research map / Graph learning / Graph self-supervised learning
Conditional Distribution Learning for Graph Classification
Jie Chen, Hua Mao, Chuanbin Liu, Zhu Wang, Xi Peng
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 pretraining (Ls) to fine-tuning (Lc+Ld) (ablation) | CDL · IMDB-B (30% labeled) | accuracy: 71.20 -> 74.60 | Improved |
| add pretraining (Ls) to fine-tuning (Lc+Ld) (ablation) | CDL · GITHUB (30% labeled) | accuracy: 68.04 -> 70.15 | Improved |
| add pretraining (Ls) to fine-tuning (Lc+Ld) (ablation) | CDL · NCI1 (70% labeled) | accuracy: 81.68 -> 82.36 | Improved |
| add conditional distribution alignment Ld to Lc+Ls (ablation) | CDL · IMDB-B (30% labeled) | accuracy: 70.70 -> 74.60 | Improved |
| add conditional distribution alignment Ld to Lc+Ls (ablation) | CDL · MUTAG (70% labeled) | accuracy: 89.39 -> 89.91 | No clear effect |
| use Ld-only fine-tuning (CDL_ft) vs Ls pretraining-only (CDL_cl) (ablation) | CDL · GITHUB (30% labeled) | accuracy: 68.44 -> 68.04 | Hurt |
| use Ld-only fine-tuning (CDL_ft) vs Ls pretraining-only (CDL_cl) (ablation) | CDL · RDT-B (50% labeled) | accuracy: 90.25 -> 91.05 | Improved |
| raise node masking ratio from 0.1 to 0.3 (ablation) | CDL · eight TU datasets | accuracy | Improved |
| raise node masking ratio to 0.35 (strong 0.7) (ablation) | CDL · eight TU datasets | accuracy | Hurt |
| CDL vs best baseline (GLIA) | CDL · MUTAG (30% labeled) | accuracy: 87.25 -> 89.36 | Improved |
Closest papers by meaning
- ReGCL: Rethinking Message Passing in Graph Contrastive Learning (AAAI 2024 · similarity 0.88)
- Graph Classification via Reference Distribution Learning: Theory and Practice (NeurIPS 2024 · similarity 0.88)
- Graph Structure Refinement with Energy-based Contrastive Learning (AAAI 2025 · similarity 0.88)
- Graph Segmentation and Contrastive Enhanced Explainer for Graph Neural Networks (AAAI 2025 · similarity 0.88)
- Hierarchical Topology Isomorphism Expertise Embedded Graph Contrastive Learning (AAAI 2024 · similarity 0.88)
- Hypergraph-enhanced Dual Semi-supervised Graph Classification (ICML 2024 · 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.
