Research map / Learning theory and algorithms / Sample complexity
Can semi-supervised learning use all the data effectively? A lower bound perspective
Alexandru Tifrea, Gizem Yüce, Amartya Sanyal, Fanny Yang
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 SSL-W (weighted SL + UL+ ensemble) instead of SL or UL+ alone | SSL-W · synthetic 2-GMM, varying SNR (nl=20, nu=2000) | classification error | Improved |
| use SSL-W instead of switching algorithm SSL-S at moderate SNR | SSL-W · synthetic 2-GMM | classification error | Improved |
| vary labeled set size at fixed SNR 0.5 and nu=7000 | SSL-W · synthetic 2-GMM, varying nu/nl | classification error | Improved |
| use SSL-W on real-world binary datasets | SSL-W · 10 OpenML and MNIST-pair datasets | classification error gap vs SL and UL+ | Improved |
| use self-training with logistic regression and ridge penalty | self-training · 10 real-world binary datasets | classification error gap vs SL and UL+ | Improved |
| self-training compared with SSL-S | self-training · 2-GMM / real data | classification error | Improved |
| use labeled data in the unsupervised step as well | UL+ · 2-GMM minimax rate | rate improvement | No clear effect |
| large unlabeled set relative to SNR | SSL · 2-GMM | improvement over UL+ | No clear effect |
| large labeled set | SSL · 2-GMM | improvement over SL | No clear effect |
Closest papers by meaning
- Semi-Supervised Sparse Gaussian Classification: Provable Benefits of Unlabeled Data (NeurIPS 2024 · similarity 0.91)
- InterLUDE: Interactions between Labeled and Unlabeled Data to Enhance Semi-Supervised Learning (ICML 2024 · similarity 0.89)
- Out-Of-Domain Unlabeled Data Improves Generalization (ICLR 2024 · similarity 0.89)
- Evaluating multiple models using labeled and unlabeled data (NeurIPS 2025 · similarity 0.88)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.88)
- Re-Evaluating the Impact of Unseen-Class Unlabeled Data on Semi-Supervised Learning Model (ICLR 2025 · 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.
