Research map / Label-efficient and robust learning / Semi-supervised learning
Newton-coupled Dual-Teacher Semi-supervised Learning Framework
Hongyang He, Xinyuan Song, Yan Zhong, Daizong Liu, Xuanyu Liu, Victor Sanchez
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 Newton-guided update (ablation) | TTN · CIFAR-10 (4k labels) | accuracy: 89.8 -> 97.5 | Improved |
| add confidence-weighted dual-teacher fusion (ablation) | TTN · CIFAR-10 (4k labels) | accuracy: 91.2 -> 97.5 | Improved |
| add MAE as second teacher to DINOv3 teacher (ablation) | TTN · CIFAR-10 (4k labels) | accuracy: 91.0 -> 97.5 | Improved |
| add augmentation consistency loss (ablation) | TTN · CIFAR-10 (4k labels) | accuracy: 92.3 -> 97.5 | Improved |
| use confidence-based Hessian fusion instead of equal fusion (ablation) | TTN · CIFAR-10 (4k labels) | accuracy: 92.0 -> 97.5 | Improved |
| use fused curvature Newton step instead of identity Hessian (ablation) | TTN · CIFAR-10 (4k labels) | accuracy: 90.8 -> 97.5 | Improved |
| add Newton-guided update (ablation) | TTN · CIFAR-10 convergence | steps to 85% accuracy: 4.8k -> 2.2k steps | Improved |
| add Newton-guided update (ablation) | TTN · CIFAR-100 | final accuracy: 80.0 -> 81.4 | Improved |
| pair MAE with DINOv3 teachers versus baseline (ablation) | TTN · ImageNet (25% labels) | accuracy gain: 3.4% improvement | Improved |
| pair MAE with DINOv2 teachers (ablation) | TTN · ImageNet (25% labels) | accuracy gain: 3.1% | Improved |
| use TTN instead of TRiCo | TTN · SVHN (1k labels) | accuracy: 94.2 -> 93.6 | Hurt |
| use TTN instead of TRiCo | TTN · ImageNet (10% labels) | accuracy: 85.9 -> 85.7 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- TRiCo — TRiCo: Triadic Game-Theoretic Co-Training for Robust Semi-Supervised Learning (NeurIPS 2025) · better on 4 of 6 shared results
- FixMatch — FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence (NeurIPS 2020) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Dual Student: Breaking the Limits of the Teacher in Semi-Supervised Learning (ICCV 2019 · similarity 0.89)
- Switching Temporary Teachers for Semi-Supervised Semantic Segmentation (NeurIPS 2023 · similarity 0.87)
- Curriculum Labeling: Revisiting Pseudo-Labeling for Semi-Supervised Learning (AAAI 2021 · similarity 0.86)
- Less or More From Teacher: Exploiting Trilateral Geometry For Knowledge Distillation (ICLR 2024 · similarity 0.85)
- DualNet: Robust Self-Supervised Stereo Matching with Pseudo-Label Supervision (AAAI 2025 · similarity 0.84)
- Revisiting Semi-Supervised Learning in the Era of Foundation Models (NeurIPS 2025 · 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.
