Research map / Label-efficient and robust learning / Semi-supervised learning
Classification from Pairwise Similarity and Unlabeled Data
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 double-hinge instead of squared loss (ablation) | SU classification · adult | accuracy: 64.5 (1.2) -> 84.5 (0.8) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · banana | accuracy: 67.5 (1.2) -> 68.2 (1.2) | Mixed |
| use squared instead of double-hinge loss (ablation) | SU classification · cod-rna | accuracy: 71.0 (0.9) -> 82.8 (1.3) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · higgs | accuracy: 55.1 (1.1) -> 69.3 (0.9) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · ijcnn1 | accuracy: 65.5 (1.3) -> 73.6 (0.9) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · magic | accuracy: 66.0 (2.0) -> 69.0 (1.3) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · phishing | accuracy: 75.0 (1.4) -> 91.3 (0.6) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · phoneme | accuracy: 67.8 (1.5) -> 70.8 (1.0) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · spambase | accuracy: 69.7 (1.4) -> 85.5 (0.5) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · susy | accuracy: 59.8 (1.3) -> 74.8 (1.2) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · w8a | accuracy: 62.1 (1.5) -> 86.5 (0.6) | Helped |
| use double-hinge instead of squared loss (ablation) | SU classification · waveform | accuracy: 77.8 (1.3) -> 87.0 (0.5) | Helped |
Closest papers by meaning
- Learning from Similarity-Confidence Data (ICML 2021 · similarity 0.89)
- Consistent Multi-Class Classification from Multiple Unlabeled Datasets (ICLR 2024 · similarity 0.88)
- AUC Optimization from Multiple Unlabeled Datasets (AAAI 2024 · similarity 0.87)
- Learning with Complementary Labels Revisited: The Selected-Completely-at-Random Setting Is More Practical (ICML 2024 · similarity 0.85)
- Scaling Up Semi-supervised Learning with Unconstrained Unlabelled Data (AAAI 2024 · similarity 0.84)
- Semi-Supervised Sparse Gaussian Classification: Provable Benefits of Unlabeled Data (NeurIPS 2024 · similarity 0.84)
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.