Research map / Graph learning / Graph self-supervised learning
Augmentation-Free Self-Supervised Learning on Graphs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace manual graph augmentations with positive-node discovery using local adjacency and global clusters | AFGRL · Node classification across WikiCS, Computers, Photo, Co.CS, and Co.Physics | accuracy | Mixed |
| Add global semantic positives to k-NN candidates (ablation) | AFGRL · Computers and WikiCS | node classification and clustering performance | Helped |
| Add local adjacency-based positives to k-NN candidates (ablation) | AFGRL · Computers and WikiCS | node classification and clustering performance | Helped |
| Use k-NN positives without adjacency filtering (ablation) | AFGRL · Computers and WikiCS | node classification and clustering performance | No clear effect |
| Use cluster positives alone, without local structural information (ablation) | AFGRL · Computers and WikiCS | node classification and clustering performance | Hurt |
| Increase embedding dimension (ablation) | AFGRL · Node-level downstream tasks | downstream performance | Helped |
| Vary the number of clustering runs M (ablation) | AFGRL · Node-level downstream tasks | downstream performance | No clear effect |
| Vary augmentation hyperparameters | BGRL and GCA · Computers, Photo, CS, and Physics; node classification and clustering | relative best-to-worst performance: -4.00%; -1.06%; -0.20%; -0.69%; -19.18%; -5.48%; -0.27%; -11.57%; -13.30%; -0.78%; -6.46%; -26.28%; -23.27%; -1.64% | Hurt |
Closest papers by meaning
- Explanation-Preserving Augmentation for Semi-Supervised Graph Representation Learning (AAAI 2026 · similarity 0.90)
- Learning the Latent Structure: A Feature-Centric Approach to Graph Data Augmentation (AAAI 2026 · similarity 0.89)
- Revisiting Self-Supervised Heterogeneous Graph Learning from Spectral Clustering Perspective (NeurIPS 2024 · similarity 0.87)
- Through the Dual-Prism: A Spectral Perspective on Graph Data Augmentation for Graph Classifications (AAAI 2025 · similarity 0.87)
- Community-Invariant Graph Contrastive Learning (ICML 2024 · similarity 0.87)
- Equivalence is All: A Unified View for Self-supervised Graph Learning (ICML 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.