Research map / Label-efficient and robust learning / Domain adaptation and generalization
Confidence Regularized Self-Training
Yang Zou, Zhiding Yu, Xiaofeng Liu, B.V.K. Vijaya Kumar, Jinsong Wang
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 MRKLD confidence regularizer | CBST · VisDA17 | mean accuracy: 76.4 -> 77.9 | Improved |
| add LRENT label regularizer | CBST · VisDA17 | mean accuracy: 76.4 -> 76.6 | Improved |
| add MRL2 regularizer | CBST · VisDA17 | mean accuracy: 76.4 -> 76.2 | No clear effect |
| add MRENT regularizer | CBST · VisDA17 | mean accuracy: 76.4 -> 76.2 | No clear effect |
| add LRENT on top of MRKLD (ablation) | MRKLD · VisDA17 | mean accuracy: 77.9 -> 78.1 | Improved |
| combine MRKLD and LRENT | CBST · Office-31 | mean accuracy: 85.8 -> 86.8 | Improved |
| add MRKLD regularizer | CBST (DeepLabv2) · GTA5 to Cityscapes | mIoU: 45.9 -> 47.1 | Improved |
| add MRKLD regularizer | CBST (ResNet-38) · GTA5 to Cityscapes | mIoU: 45.2 -> 46.8 | Improved |
| add MRKLD regularizer with spatial priors | CBST-SP (ResNet-38) · GTA5 to Cityscapes | mIoU: 48.1 -> 49.2 | Improved |
| add multi-scale testing (ablation) | MRKLD-SP (ResNet-38) · GTA5 to Cityscapes | mIoU: 49.2 -> 49.8 | Improved |
| add MRKLD regularizer | CBST · SYNTHIA to Cityscapes | mIoU: 42.6 -> 43.8 | Improved |
| vary starting portion p0 from 20 to 15 (ablation) | MRL2 · Office-31 W to A | accuracy: 72.1 -> 71.3 | No clear effect |
Closest papers by meaning
- Theoretical Analysis of Self-Training with Deep Networks on Unlabeled Data (ICLR 2021 · similarity 0.89)
- Neighborhood-Regularized Self-Training for Learning with Few Labels (AAAI 2023 · similarity 0.88)
- Instance Adaptive Self-Training for Unsupervised Domain Adaptation (ECCV 2020 · similarity 0.88)
- Cycle Self-Training for Domain Adaptation (NeurIPS 2021 · similarity 0.88)
- Pseudo-Labeling for Unsupervised Domain Adaptation with Kernel GLMs (NeurIPS 2026 · similarity 0.86)
- Toward Improving Robustness and Accuracy in Unsupervised Domain Adaptation (AAAI 2025 · similarity 0.86)
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.
