Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-Supervised Knowledge Amalgamation for Sequence Classification
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Teacher Coordinator to estimate teacher trust and rescale teacher probabilities | Sequence-classification student trained from teachers · SyntheticControl, 2% labeled | accuracy: .87±.01 -> .90±.02 | Helped |
| Use Teacher Coordinator to estimate teacher trust and rescale teacher probabilities | Sequence-classification student trained from teachers · MelbournePedestrian, 2% labeled | accuracy: .61±.03 -> .69±.01 | Helped |
| Use Teacher Coordinator to estimate teacher trust and rescale teacher probabilities | Sequence-classification student trained from teachers · Human Activity Recognition, 2% labeled | accuracy: .66±.10 -> .75±.01 | Helped |
| Use Teacher Coordinator to estimate teacher trust and rescale teacher probabilities | Sequence-classification student trained from teachers · ElectricDevices, 2% labeled | accuracy: .65±.01 -> .66±.01 | Helped |
| Use Teacher Coordinator instead of the strongest reported alternative | Student learning from teachers with unequal class-set sizes · SyntheticControl, overconfident teachers | accuracy: .66±.01 -> .85±.07 | Helped |
| Use Teacher Coordinator instead of the strongest reported alternative | Student learning from teachers with unequal class-set sizes · MelbournePedestrian, overconfident teachers | accuracy: .66±.02 -> .70±.08 | Helped |
| Use Teacher Coordinator instead of the strongest reported alternative | Student learning from teachers with unequal class-set sizes · Human Activity Recognition, overconfident teachers | accuracy: .58±.00 -> .75±.01 | Helped |
| Use Teacher Coordinator instead of the strongest reported alternative | Student learning from teachers with unequal class-set sizes · ElectricDevices, overconfident teachers | accuracy: .44±.00 -> .64±.01 | Helped |
| Combine disjoint teacher class sets using TC's trust-based probability rescaling | Student amalgamating disparate teachers · SyntheticControl, exclusive class sets | accuracy: .61±.02 -> .77±.07 | Helped |
| Combine disjoint teacher class sets using TC's trust-based probability rescaling | Student amalgamating disparate teachers · MelbournePedestrian, exclusive class sets | accuracy: .48±.02 -> .64±.04 | Helped |
| Combine disjoint teacher class sets using TC's trust-based probability rescaling | Student amalgamating disparate teachers · Human Activity Recognition, exclusive class sets | accuracy: .55±.02 -> .84±.04 | Helped |
| Increase the proportion of student-training data with labels | Teacher Coordinator · SyntheticControl, labeled-data proportions from 2% to 8% | accuracy: .90±.02 -> .93±.01 | Helped |
Closest papers by meaning
- DistiLLM: Towards Streamlined Distillation for Large Language Models (ICML 2024 · similarity 0.83)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.83)
- MiniPLM: Knowledge Distillation for Pre-training Language Models (ICLR 2025 · similarity 0.83)
- Flipped Classroom: Aligning Teacher Attention with Student in Generalized Category Discovery (NeurIPS 2024 · similarity 0.83)
- Targeted Representation Alignment for Open-World Semi-Supervised Learning (CVPR 2024 · similarity 0.83)
- Debiased Distillation for Consistency Regularization (AAAI 2025 · similarity 0.82)
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.