Research map / Label-efficient and robust learning / Knowledge distillation
Meta-prediction Model for Distillation-Aware NAS on Unseen Datasets
Hayeon Lee, Sohyun An, Minseon Kim, Sung Ju Hwang
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 |
|---|---|---|---|
| encode the teacher and parameter-remapped student function instead of conditioning only on architecture or dataset | DaSS · Unseen DaNAS tasks (CUB, Cars, DTD, Quick Draw) | Spearman's rank correlation coefficient: 0.45 -> 0.64 | Improved |
| add teacher-accuracy-guided adaptation (ablation) | DaSS · CUB | Spearman's rank correlation coefficient: 0.69 -> 0.69 | No clear effect |
| add teacher-accuracy-guided adaptation (ablation) | DaSS · Stanford Cars | Spearman's rank correlation coefficient: 0.77 -> 0.88 | Improved |
| add teacher-accuracy-guided adaptation (ablation) | DaSS · DTD | Spearman's rank correlation coefficient: 0.50 -> 0.57 | Improved |
| add teacher-accuracy-guided adaptation (ablation) | DaSS · Quick Draw | Spearman's rank correlation coefficient: 0.62 -> 0.75 | Improved |
| add teacher-accuracy-guided adaptation (ablation) | DaSS · Mean across CUB, Cars, DTD, and Quick Draw | Spearman's rank correlation coefficient: 0.64 -> 0.72 | Improved |
| use DaSS to select the student architecture rather than the strongest meta-NAS baseline | DaNAS · CUB | accuracy (%): 41.74 ± 0.34 -> 43.48 ± 0.11 | Improved |
| use DaSS to select the student architecture rather than the strongest meta-NAS baseline | DaNAS · Stanford Cars | accuracy (%): 56.74 ± 0.36 -> 59.09 ± 0.06 | Improved |
| use DaSS to select the student architecture rather than the strongest meta-NAS baseline | DaNAS · DTD | accuracy (%): 45.26 ± 0.78 -> 47.86 ± 0.40 | Improved |
| use DaSS to select the student architecture rather than the strongest meta-NAS baseline | DaNAS · Quick Draw | accuracy (%): 46.80 ± 0.09 -> 47.27 ± 0.10 | Improved |
| use DaSS to select the student architecture rather than TANS | DaNAS · CropDisease | accuracy (%): 99.72 ± 0.02 -> 99.71 ± 0.01 | No clear effect |
| use DaSS to select the student architecture rather than TANS | DaNAS · EuroSAT | accuracy (%): 97.33 ± 0.48 -> 97.95 ± 0.06 | Improved |
Closest papers by meaning
- Performance-Aware Mutual Knowledge Distillation for Improving Neural Architecture Search (CVPR 2022 · similarity 0.90)
- MNGNAS: Distilling Adaptive Combination of Multiple Searched Networks for One-Shot Neural Architecture Search (TPAMI 2023 · similarity 0.90)
- RNAS-CL: Robust Neural Architecture Search by Cross-Layer Knowledge Distillation (IJCV 2024 · similarity 0.90)
- Search to Distill: Pearls Are Everywhere but Not the Eyes (CVPR 2020 · similarity 0.89)
- DisWOT: Student Architecture Search for Distillation WithOut Training (CVPR 2023 · similarity 0.89)
- M-NAS: Meta Neural Architecture Search (AAAI 2020 · similarity 0.89)
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.
