Research map / Architectures and efficiency / Neural architecture search
Block-Wisely Supervised Neural Architecture Search With Knowledge Distillation
Check 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 block-wise architecture search with teacher feature supervision instead of one-shot shared-weight ranking | NAS on ImageNet · ImageNet model-ranking test | correlation between predicted and stand-alone accuracy | Helped |
| Use DNA search rather than EfficientNet-B0 | ImageNet mobile models · ImageNet val, 224×224 | Acc@1: 76.3% -> 77.5% | Helped |
| Use DNA-c rather than EfficientNet-B0 | transfer learning · CIFAR-10 | accuracy: 98.0% -> 98.3% | Helped |
| Use DNA-c rather than EfficientNet-B0 | transfer learning · CIFAR-100 | accuracy: 87.1% -> 88.3% | Helped |
| Replace progressive S1/S2 training with teacher-previous-feature inputs for each block (ablation) | DNA architecture distillation · ImageNet | Acc@1: 77.0% -> 77.49%; 77.15% -> 77.49% | Helped |
| Add multi-cell channel and layer variability (ablation) | DNA search without constraint · ImageNet | Acc@1: 77.49% -> 77.84% | Helped |
| Add multi-cell channel and layer variability (ablation) | DNA search under a model-size constraint · ImageNet | Acc@1: 77.21% -> 77.38% | Helped |
| Use EfficientNet-B0 instead of EfficientNet-B7 as the teacher (ablation) | DNA architecture search · ImageNet | Acc@1: 77.8% -> 77.8% | No clear effect |
| Scale DNA-B7 to a similar model size as its teacher (ablation) | DNA-B7 · ImageNet | Acc@1: 77.8% -> 79.9% | Helped |
Closest papers by meaning
- Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective (ICLR 2021 · similarity 0.92)
- Understanding Architectures Learnt by Cell-based Neural Architecture Search (ICLR 2020 · similarity 0.92)
- NAS-Bench-201: Extending the Scope of Reproducible Neural Architecture Search (ICLR 2020 · similarity 0.91)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.90)
- einspace: Searching for Neural Architectures from Fundamental Operations (NeurIPS 2024 · similarity 0.90)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.90)
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.