Research map / Graph learning / Graph self-supervised learning
Graph-Wise Common Latent Factor Extraction for Unsupervised Graph Representation Learning
Thilini Cooray, Ngai-Man Cheung
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 deepGCFX++ common-plus-local graph embedding instead of deepGCFX common-only embedding | deepGCFX · Graph classification across six datasets | accuracy | Improved |
| Use deepGCFX++ common-plus-local node representations instead of local factors alone | deepGCFX · Node classification across Cora, Citeseer, Pubmed, Chameleon, Squirrel, and Actor | accuracy | Improved |
| Incorporate graph-wise common factors through deepGCFX++ | node classification on disassortative graphs · Chameleon, Squirrel, and Actor | accuracy: 61.05 vs 59.45; 39.20 vs 36.33; 28.80 vs 27.09 | Improved |
| Use 0 instead of 2 ACCUM iterations (ablation) | deepGCFX++ · MUTAG | accuracy: 92.2 ± 0.9(0.7) -> 87.2 ± 1.2(0.1) | Hurt |
| Use 1 instead of 2 ACCUM iterations (ablation) | deepGCFX++ · MUTAG | accuracy: 92.2 ± 0.9(0.7) -> 89.1 ± 0.9(0.4) | Hurt |
| Use 3 instead of 2 ACCUM iterations (ablation) | deepGCFX++ · MUTAG | accuracy: 92.2 ± 0.9(0.7) -> 90.5 ± 1.2(0.8) | Hurt |
| Use 4 instead of 2 ACCUM iterations (ablation) | deepGCFX++ · MUTAG | accuracy: 92.2 ± 0.9(0.7) -> 89.3 ± 0.9(0.4) | Hurt |
| Use 0 instead of 3 ACCUM iterations (ablation) | deepGCFX++ · IMDB-MULTI | accuracy: 52.7 ± 0.4(0.85) -> 48.8 ± 0.5(0.0) | Hurt |
| Use 1 instead of 3 ACCUM iterations (ablation) | deepGCFX++ · IMDB-MULTI | accuracy: 52.7 ± 0.4(0.85) -> 49.4± 0.3(0.3) | Hurt |
| Use 2 instead of 3 ACCUM iterations (ablation) | deepGCFX++ · IMDB-MULTI | accuracy: 52.7 ± 0.4(0.85) -> 51.2 ± 0.4(0.7) | Hurt |
| Use 4 instead of 3 ACCUM iterations (ablation) | deepGCFX++ · IMDB-MULTI | accuracy: 52.7 ± 0.4(0.85) -> 50.8 ± 0.5(0.7) | Hurt |
| Increase ACCUM iterations from random initialization to learned common-factor extraction (ablation) | deepGCFX++ · MUTAG and IMDB-MULTI graph classification | accuracy: 87.2 ± 1.2(0.1) -> 92.2 ± 0.9(0.7); 48.8 ± 0.5(0.0) -> 52.7 ± 0.4(0.85) | Mixed |
Closest papers by meaning
- InfoGraph: Unsupervised and Semi-supervised Graph-Level Representation Learning via Mutual Information Maximization (ICLR 2020 · similarity 0.88)
- Representing Graphs via Gromov-Wasserstein Factorization (TPAMI 2022 · similarity 0.87)
- Learning the Latent Structure: A Feature-Centric Approach to Graph Data Augmentation (AAAI 2026 · similarity 0.86)
- Maximum Common Subgraph Guided Graph Retrieval: Late and Early Interaction Networks (NeurIPS 2022 · similarity 0.86)
- Aitchison Embeddings for Learning Compositional Graph Representations (ICML 2026 · similarity 0.86)
- Pre-Training Graph Neural Networks on Molecules by Using Subgraph-Conditioned Graph Information Bottleneck (AAAI 2025 · similarity 0.85)
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.
