Research map / Graph learning / Graph algorithms and clustering
Implicit SVD for Graph Representation Learning
Sami Abu-El-Haija, Hesham Mostafa, Marcel Nassar, Valentino Crespi, Greg Ver Steeg, Aram Galstyan
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 pseudo-dropout augmentation to the design matrix (ablation) | iSVD₁₀₀ · Cora | accuracy: 82.0 { $\pm$0.13} -> 82.5 { $\pm$0.46} | No clear effect |
| Add pseudo-dropout augmentation to the design matrix (ablation) | iSVD₁₀₀ · Citeseer | accuracy: 71.4 { $\pm$0.22} -> 71.5 { $\pm$0.53} | No clear effect |
| Add pseudo-dropout augmentation to the design matrix (ablation) | iSVD₁₀₀ · Pubmed | accuracy: 78.9 { $\pm$0.31} -> 78.9 { $\pm$0.59} | No clear effect |
| Add one-epoch finetuning of f(μ,s) (ablation) | iSVD₁₀₀ · ogbl-DDI | Hits@20: 67.86 { $\pm$0.09} -> 79.09 { $\pm$0.18} | Improved |
| Update the SVD basis with validation edges while keeping f(μ,s) fixed (ablation) | iSVD₁₀₀ + finetune f(μ,s) · ogbl-DDI | Hits@20: 79.09 { $\pm$0.18} -> 84.09 { $\pm$0.03} | Improved |
| Add a dropout replica of the label-reuse features (ablation) | iSVD₂₅₀ · ogbn-ArXiv | accuracy: 68.90 { $\pm$0.02} -> 69.34 { $\pm$0.02} | Improved |
| Add two dropout replicas of the design matrix with different seeds (ablation) | iSVD₂₅₀ + dropout(LR) · ogbn-ArXiv | accuracy: 69.34 { $\pm$0.02} -> 71.95 { $\pm$0.03} | Improved |
| Fine-tune the initialized network for 15 epochs (ablation) | iSVD₂₅₀ + dropout(M) · ogbn-ArXiv | accuracy: 71.95 { $\pm$0.03} -> 74.14 { $\pm$0.05} | Improved |
| Append the network-embedding decomposition to node features (ablation) | ogbn-ArXiv classification model · ogbn-ArXiv | validation performance | No clear effect |
Closest papers by meaning
- Graph Traversal with Tensor Functionals: A Meta-Algorithm for Scalable Learning (ICLR 2021 · similarity 0.88)
- OpenGSL: A Comprehensive Benchmark for Graph Structure Learning (NeurIPS 2023 · similarity 0.86)
- Exploitation of a Latent Mechanism in Graph Contrastive Learning: Representation Scattering (NeurIPS 2024 · similarity 0.86)
- Learning Representations for Hierarchies with Minimal Support (NeurIPS 2024 · similarity 0.86)
- S3GCL: Spectral, Swift, Spatial Graph Contrastive Learning (ICML 2024 · similarity 0.86)
- Graph Contrastive Invariant Learning from the Causal Perspective (AAAI 2024 · 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.
