Research map / Graph learning / Graph self-supervised learning
Lovász Principle for Unsupervised Graph Representation Learning
Ziheng Sun, Chris Ding, Jicong Fan
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 |
|---|---|---|---|
| Replace InfoMax loss with Lovasz loss | InfoGraph · MUTAG unsupervised | accuracy: 89.01 -> 89.67 | Improved |
| Replace InfoMax loss with Lovasz loss | GraphCL, AD-GCL, JOAOv2, AutoGCL · 8 TUD datasets unsupervised | accuracy: 86.80 -> 87.24 | Improved |
| Use enhanced Lovasz loss instead of Lovasz loss (add subgraph Lovasz number loss) (ablation) | InfoGraph · NCI1 unsupervised | accuracy: 78.21 -> 79.36 | Improved |
| Add subgraph Lovasz number loss (eta>0) (ablation) | Lovasz loss with GraphCL · COLLAB unsupervised | accuracy: 72.52 -> 74.16 | Improved |
| Add subgraph Lovasz number loss (eta>0) (ablation) | Lovasz loss with InfoGraph · REDDIT-B unsupervised | accuracy: 84.76 -> 84.53 | Hurt |
| Add orthonormal representation regularization (mu>0) (ablation) | Lovasz loss with InfoGraph · MUTAG unsupervised | accuracy: 88.25 -> 89.67 | Improved |
| Add orthonormal representation regularization (mu>0) (ablation) | Lovasz loss with JOAOv2 · MUTAG unsupervised | accuracy: 88.34 -> 87.19 | Hurt |
| Use constrained optimization instead of regularized optimization (ablation) | Lovasz loss with InfoGraph · MUTAG unsupervised | accuracy: 89.67 -> 86.12 | Hurt |
| Use constrained optimization instead of regularized optimization (ablation) | Lovasz loss with GraphCL · PROTEINS unsupervised | accuracy: 75.87 -> 76.11 | No clear effect |
| Increase gamma (orthogonal regularization weight) very large (ablation) | Lovasz principle · 8 datasets average ACC | accuracy | Hurt |
| Increase eta (SLN loss weight) very large (ablation) | enhanced Lovasz loss · 8 datasets average ACC | accuracy | Hurt |
| Replace InfoMax with Lovasz loss in semi-supervised training | AutoGCL · NCI1 semi-supervised | accuracy: 73.75 -> 75.77 | Improved |
Closest papers by meaning
- Representing Graphs via Gromov-Wasserstein Factorization (TPAMI 2022 · similarity 0.86)
- Wasserstein Embedding for Graph Learning (ICLR 2021 · similarity 0.86)
- Learning Graph Representation via Graph Entropy Maximization (ICML 2024 · similarity 0.85)
- OpenGSL: A Comprehensive Benchmark for Graph Structure Learning (NeurIPS 2023 · similarity 0.85)
- Unsupervised Adversarially Robust Representation Learning on Graphs (AAAI 2022 · similarity 0.85)
- GrokFormer: Graph Fourier Kolmogorov-Arnold Transformers (ICML 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.
