Research map / Label-efficient and robust learning / Semi-supervised learning
FisherMatch: Semi-Supervised Rotation Regression via Entropy-Based Filtering
Yingda Yin, Yingcheng Cai, He Wang, Baoquan Chen
Check 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 entropy-based pseudo-label filtering to teacher-student SSL | Rotation regression with FisherMatch · ModelNet10-SO(3) Sofa, 5% labeled | Mean error (degrees): 36.86 -> 32.02 | Improved |
| Add entropy-based pseudo-label filtering to teacher-student SSL | Rotation regression with FisherMatch · ModelNet10-SO(3) Sofa, 10% labeled | Mean error (degrees): 25.94 -> 21.29 | Improved |
| Add entropy-based pseudo-label filtering to teacher-student SSL | Rotation regression with FisherMatch · ModelNet10-SO(3) Chair, 5% labeled | Mean error (degrees): 31.20 -> 26.69 | Improved |
| Add entropy-based pseudo-label filtering to teacher-student SSL | Rotation regression with FisherMatch · ModelNet10-SO(3) Chair, 10% labeled | Mean error (degrees): 23.59 -> 20.06 | Improved |
| Use a Bingham distribution and quaternion representation instead of matrix Fisher (ablation) | Semi-supervised rotation regression · ModelNet10-SO(3) Sofa, 10% labeled | Mean error (degrees): 21.29 -> 27.01 | Hurt |
| Use a Bingham distribution and quaternion representation instead of matrix Fisher (ablation) | Semi-supervised rotation regression · ModelNet10-SO(3) Sofa, 10% labeled | Median error (degrees): 5.25 -> 6.77 | Hurt |
| Replace cross-entropy consistency loss with NLL on the teacher's predicted mode (ablation) | FisherMatch · ModelNet10-SO(3) Sofa, 5% labeled | Rotation error | No clear effect |
| Use a more tolerant entropy threshold with cross-entropy loss (ablation) | FisherMatch · ModelNet10-SO(3) Sofa, 5% labeled | Rotation error | Mixed |
Closest papers by meaning
- Learning with 3D rotations, a hitchhiker's guide to SO(3) (ICML 2024 · similarity 0.84)
- Semi-supervised Class-Agnostic Motion Prediction with Pseudo Label Regeneration and BEVMix (AAAI 2024 · similarity 0.84)
- Non-Stationary Predictions May Be More Informative: Exploring Pseudo-Labels with a Two-Phase Pattern of Training Dynamics (ICML 2025 · similarity 0.84)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.84)
- Semi-Supervised Regression with Heteroscedastic Pseudo-Labels (NeurIPS 2025 · similarity 0.83)
- Revisiting Semi-Supervised Learning in the Era of Foundation Models (NeurIPS 2025 · similarity 0.83)
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.
