Research map / Graph learning / Graph self-supervised learning
Consistency of Augmentation Graph and Network Approximability in Contrastive Learning
Chenghui Li, A. Martina Neuman
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 theory-based noise parameter spectral method instead of SimCLR | spectral contrastive learning · CIFAR-10, 200 epochs | top-1 accuracy: 83.73 -> 87.80 | Improved |
| use theory-based noise spectral method instead of SimCLR | spectral contrastive learning · CIFAR-100, 800 epochs | top-1 accuracy: 63.88 -> 66.36 | Improved |
| use theory-based noise spectral method instead of SimSiam | spectral contrastive learning · CIFAR-100, 800 epochs | top-1 accuracy: 65.87 -> 66.36 | Improved |
| use theory-based noise spectral method instead of SimSiam | spectral contrastive learning · CIFAR-10, 400 epochs | top-1 accuracy: 90.31 -> 89.79 | Hurt |
| use theory-based noise spectral method instead of SimSiam | spectral contrastive learning · CIFAR-100, 200 epochs | top-1 accuracy: 61.56 -> 58.64 | Hurt |
| use theory-based noise parameter instead of tuned prior spectral contrastive implementation | spectral contrastive learning · CIFAR-100, 200 epochs | top-1 accuracy: 62.45 -> 58.64 | Hurt |
| use theory-based noise parameter instead of tuned prior spectral contrastive implementation | spectral contrastive learning · CIFAR-100, 800 epochs | top-1 accuracy: 66.18 -> 66.36 | Improved |
| use theory-based noise parameter instead of tuned prior spectral contrastive implementation | spectral contrastive learning · CIFAR-10, 800 epochs | top-1 accuracy: 92.07 -> 91.70 | Hurt |
Closest papers by meaning
- Spectral Feature Augmentation for Graph Contrastive Learning and Beyond (AAAI 2023 · similarity 0.91)
- Directed Graph Contrastive Learning (NeurIPS 2021 · similarity 0.90)
- Perfect Alignment May be Poisonous to Graph Contrastive Learning (ICML 2024 · similarity 0.89)
- Community-Invariant Graph Contrastive Learning (ICML 2024 · similarity 0.89)
- A Statistical Theory of Contrastive Learning via Approximate Sufficient Statistics (NeurIPS 2025 · similarity 0.87)
- Contrastive Learning is Spectral Clustering on Similarity Graph (ICLR 2024 · 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.
