Research map / Graph learning / Graph self-supervised learning
Coarse-to-Fine Open-Set Graph Node Classification with Large Language Models
Xueqi Ma, Xingjun Ma, Sarah Monazam Erfani, Danilo Mandic, James Bailey
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 manifold mixup OOD augmentation (with denoising) over neither (ablation) | CFC · Cora (2 OOD classes) | overall accuracy: 88.20 -> 90.00 | Improved |
| add denoising via label propagation (ablation) | CFC · Cora (2 OOD classes) | ID accuracy: 85.44 -> 88.63 | Improved |
| add manifold mixup augmentation (ablation) | CFC · Citeseer (2 OOD classes) | overall accuracy: 73.44 -> 77.21 | Improved |
| add denoising (ablation) | CFC · Citeseer (2 OOD classes) | overall accuracy: 77.65 -> 77.21 | Hurt |
| use semantic OOD samples from LLM without denoising or augmentation versus baselines (ablation) | CFC (wo / D/M) · Cora (2 OOD classes) | overall accuracy: 85.71 -> 88.20 | Improved |
| add confidence constraint (reject only when highly confident) / candidate OOD label space in LLM prompt (ablation) | coarse classifier · Cora and Citeseer | OOD detection AUROC | Improved |
| use GPT-4o instead of Llama2-7b or Llama3-8b for coarse OOD detection (ablation) | coarse classifier · Cora and Citeseer | OOD detection | Improved |
| add post-OOD label space to LLM prompt for OOD classification (ablation) | CFC OOD classification · Cora and Citeseer | OOD classification accuracy | Improved |
| increase number of OOD samples identified by coarse classifier (ablation) | CFC · Cora and Citeseer | accuracy | Improved |
| increase number of manifold-mixup generated OOD samples (ablation) | CFC · Cora and Citeseer | accuracy | Mixed |
| use CFC instead of closed-set GCN_sigmoid | GCN · Cora (2 OOD classes) | ID accuracy: 90.64 -> 87.49 | Hurt |
Closest papers by meaning
- GOODAT: Towards Test-Time Graph Out-of-Distribution Detection (AAAI 2024 · similarity 0.88)
- GOLD: Graph Out-of-Distribution Detection via Implicit Adversarial Latent Generation (ICLR 2025 · similarity 0.88)
- ROG_PL: Robust Open-Set Graph Learning via Region-Based Prototype Learning (AAAI 2024 · similarity 0.88)
- Structural Entropy Guided Unsupervised Graph Out-Of-Distribution Detection (AAAI 2025 · similarity 0.88)
- ML-GOOD: Towards Multi-Label Graph Out-Of-Distribution Detection (AAAI 2025 · similarity 0.88)
- Harnessing Feature Resonance under Arbitrary Target Alignment for Out-of-Distribution Node Detection (NeurIPS 2025 · similarity 0.88)
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.
