Research map / Graph learning / Graph self-supervised learning
Randomized Schur Complement Views for Graph Contrastive Learning
Vignesh Kothapalli
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 randomized Schur-complement views instead of competing augmentors | GRACE design · CORA, Amazon-Photo, PubMed, Coauthor-CS, Coauthor-Phy | accuracy: 83.75 ± 2.64; 92.59 ± 1.05; 84.56 ± 0.71; 93.1 ± 0.54; 95.83 ± 0.44 | Improved |
| use randomized Schur-complement views instead of competing augmentors | MVGRL design · CORA and Amazon-Photo | accuracy: 83.68 ± 2.04; 87.14 ± 1.34 | Hurt |
| use randomized Schur-complement views instead of competing augmentors | MVGRL design · PubMed, Coauthor-CS, Coauthor-Phy | accuracy: 84.21 ± 0.46; 91.73 ± 0.53; 94.81 ± 0.31 | Improved |
| use randomized Schur-complement views instead of competing augmentors | GraphCL design · PROTEINS, MUTAG, IMDB-MULTI, NCI1 | accuracy: 75.27 ± 3.34; 87.5 ± 9.86; 48.66 ± 2.68; 75.06 ± 1.65 | Improved |
| use randomized Schur-complement views instead of competing augmentors | GraphCL design · IMDB-BINARY | accuracy: 70.9 ± 5.01; 71.4 ± 2.87 | Hurt |
| use randomized Schur-complement views instead of competing augmentors | BGRL design · PROTEINS, MUTAG, IMDB-MULTI, NCI1 | accuracy: 84.34 ± 4.03; 81.5 ± 5.39; 59.47 ± 7.42; 75.15 ± 3.57 | Improved |
| use randomized Schur-complement views instead of competing augmentors | BGRL design · IMDB-BINARY | accuracy: 74.8 ± 10.17; 78.5 ± 6.32 | Hurt |
| use minimum-degree instead of random node elimination (ablation) | rLap with BGRL design · IMDB-BINARY | accuracy: ≈ 79.5% | Improved |
| combine stochasticity with topology-preserving Schur-complement elimination | rLap · node and graph classification benchmarks | accuracy | Mixed |
Closest papers by meaning
- Simple and Asymmetric Graph Contrastive Learning without Augmentations (NeurIPS 2023 · similarity 0.91)
- Perfect Alignment May be Poisonous to Graph Contrastive Learning (ICML 2024 · similarity 0.90)
- Community-Invariant Graph Contrastive Learning (ICML 2024 · similarity 0.90)
- Hierarchical Topology Isomorphism Expertise Embedded Graph Contrastive Learning (AAAI 2024 · similarity 0.89)
- TopoGCL: Topological Graph Contrastive Learning (AAAI 2024 · similarity 0.89)
- S3GCL: Spectral, Swift, Spatial Graph Contrastive Learning (ICML 2024 · 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.
