Research map / Architectures and efficiency / Neural architecture search
NAT: Neural Architecture Transformer for Accurate and Compact Architectures
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 |
|---|---|---|---|
| Optimize VGG16 with NAT | VGG16 · CIFAR-10 | accuracy (%): 93.56 -> 96.04 | Helped |
| Optimize ResNet20 with NAT | ResNet20 · CIFAR-10 | accuracy (%): 91.37 -> 92.95 | Helped |
| Optimize MobileNetV2 with NAT | MobileNetV2 · CIFAR-10 | accuracy (%): 94.47 -> 95.17 | Helped |
| Optimize VGG16 with NAT | VGG16 · ImageNet | Top-1 accuracy (%): 71.6 -> 74.3 | Helped |
| Optimize ResNet18 with NAT | ResNet18 · ImageNet | Top-1 accuracy (%): 69.8 -> 71.1 | Helped |
| Optimize MobileNetV2 with NAT | MobileNetV2 · ImageNet | Top-1 accuracy (%): 72.0 -> 72.5 | Helped |
| Optimize DARTS with NAT | DARTS · CIFAR-10 | accuracy (%): 97.06 -> 97.28 | Helped |
| Optimize DARTS with NAT | DARTS · ImageNet | Top-1 accuracy (%): 73.1 -> 73.7 | Helped |
| Optimize NAONet with NAT | NAONet · CIFAR-10 | accuracy (%): 97.89 -> 98.01 | Helped |
| Optimize NAONet with NAT | NAONet · ImageNet | Top-1 accuracy (%): 74.3 -> 74.8 | Helped |
| Use sampling from the learned GCN policy rather than maximum-probability selection (ablation) | GCN policy learner · CIFAR-10 DARTS | accuracy (%): 97.00 -> 97.26 | Helped |
| Replace the GCN policy learner with LSTM (ablation) | NAT · CIFAR-10 DARTS | accuracy (%): 97.05 -> 97.26 | Hurt |
Closest papers by meaning
- Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective (ICLR 2021 · similarity 0.87)
- Evaluating The Search Phase of Neural Architecture Search (ICLR 2020 · similarity 0.87)
- Neural Architecture Generator Optimization (NeurIPS 2020 · similarity 0.87)
- Transfer NAS with Meta-learned Bayesian Surrogates (ICLR 2023 · similarity 0.87)
- NADER: Neural Architecture Design via Multi-Agent Collaboration (CVPR 2025 · similarity 0.87)
- Robustifying and Boosting Training-Free Neural Architecture Search (ICLR 2024 · similarity 0.86)
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.