Research map / Label-efficient and robust learning / Self-supervised and contrastive
Quantitative Estimation of Target Task Performance from Unsupervised Pretext Task in Semi/Self-Supervised Learning
Lin-Han Jia, Siyu Han, Wen-Chao Hu, Jie-Jing Shao, Wen-Da Wei, Zhi Zhou, Lan-Zhe Guo, Yu-Feng Li
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 reliability estimate to learnability alone (ablation) | Performance estimator · CIFAR-10 self-supervised | Pearson correlation: 0.224 -> 0.716 | Improved |
| add completeness estimate to learnability alone (ablation) | Performance estimator · CIFAR-10 self-supervised | Pearson correlation: 0.224 -> 0.746 | Improved |
| add learnability to reliability + completeness (ablation) | Performance estimator · CIFAR-10 self-supervised | Pearson correlation: 0.808 -> 0.820 | Improved |
| add learnability to reliability + completeness (ablation) | Performance estimator · CIFAR-10 semi-supervised | Pearson correlation: 0.773 -> 0.785 | Improved |
| use all three indicators instead of any single one (ablation) | Performance estimator · CIFAR-10 semi-supervised | Pearson correlation: 0.692 -> 0.785 | Improved |
| increase unlabeled estimation data from 10x to 50x the labeled data (ablation) | Performance estimator · CIFAR-10 self-supervised | Pearson correlation: 0.820 -> 0.854 | Improved |
| increase unlabeled estimation data from 10x to 50x the labeled data (ablation) | Performance estimator · CIFAR-10 semi-supervised | Pearson correlation: 0.785 -> 0.803 | Improved |
| train longer (10 epochs instead of 5) (ablation) | Performance estimator · CIFAR-10 semi-supervised | Pearson correlation: 0.785 -> 0.874 | Improved |
| lower learning rate to 1e-5 (ablation) | Performance estimator · CIFAR-10 semi-supervised | Pearson correlation: 0.785 -> 0.666 | Hurt |
| swap ViT-B backbone for ResNet-50 (ablation) | Performance estimator · CIFAR-10 semi-supervised | Pearson correlation: 0.785 -> 0.478 | Hurt |
| use estimator instead of full training and validation | Pretext task assessment · CIFAR-10 | time (s): 1651.6 -> 29.6 | Improved |
| use contrastive/semi-supervised algorithm-specific training (e.g. MoCo) instead of the basic unified algorithm (ablation) | Performance estimator · CIFAR-10 self-supervised | Pearson correlation: 0.820 -> 0.453 | Hurt |
Closest papers by meaning
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.87)
- Realistic Evaluation of Deep Semi-Supervised Learning Algorithms (NeurIPS 2018 · similarity 0.87)
- Re-Evaluating the Impact of Unseen-Class Unlabeled Data on Semi-Supervised Learning Model (ICLR 2025 · similarity 0.86)
- On the Provable Advantage of Unsupervised Pretraining (ICLR 2024 · similarity 0.86)
- Rethinking Confidence Scores and Thresholds in Pseudolabeling-based SSL (ICML 2025 · similarity 0.86)
- Revisiting Semi-Supervised Learning in the Era of Foundation Models (NeurIPS 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.
