Research map / Architectures and efficiency / Neural architecture search
Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective
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 |
|---|---|---|---|
| combine NTK condition-number and linear-region indicators when pruning (ablation) | TE-NAS on NAS-Bench-201 · CIFAR-100 | test accuracy: 69.25 (1.29); 70.48 (0.29) -> 71.24 (0.56) | Helped |
| sum the two indicator rankings rather than use their minimum (ablation) | TE-NAS on NAS-Bench-201 · CIFAR-100 | test accuracy: 70.75 (0.73) -> 71.24 (0.56) | Helped |
| sum the two indicator rankings rather than use their maximum (ablation) | TE-NAS on NAS-Bench-201 · CIFAR-100 | test accuracy: 70.33 (1.09) -> 71.24 (0.56) | Helped |
| rank indicator changes before combining them instead of summing numerical changes (ablation) | TE-NAS on NAS-Bench-201 · CIFAR-100 | test accuracy: 70.47 (0.68) -> 71.24 (0.56) | Helped |
| use equal-weight ranking summation instead of other tested weighted summations (ablation) | TE-NAS · NAS search | test accuracy | Helped |
| prune operators using both trainability and expressivity indicators | TE-NAS on DARTS · ImageNet mobile setting, architecture searched on ImageNet | top-1 test error (%): 24.2; 24.9 -> 24.5 | Mixed |
Closest papers by meaning
- NetAdaptV2: Efficient Neural Architecture Search With Fast Super-Network Training and Architecture Optimization (CVPR 2021 · similarity 0.93)
- Vision-Oriented Lightweight Neural Architecture Search with Budget-Adaptive Evaluation (CVPR 2026 · similarity 0.93)
- Evaluating The Search Phase of Neural Architecture Search (ICLR 2020 · similarity 0.93)
- einspace: Searching for Neural Architectures from Fundamental Operations (NeurIPS 2024 · similarity 0.92)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.92)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.91)
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.