Research map / Architectures and efficiency / Neural architecture search
Few-Shot Neural Architecture Search
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 |
|---|---|---|---|
| Replace one supernet with multiple sub-supernets (ablation) | NAS methods on NasBench-201 · NasBench-201 | Kendall’s Tau: 0.5436 -> 0.653; 0.696; 0.752 | Helped |
| Split one rather than multiple compound edges (ablation) | Few-shot NAS split strategy · NasBench-201 | Kendall’s Tau: 0.653; 0.696; 0.752 | Mixed |
| Choose different edges to split while keeping the number of edges fixed (ablation) | Few-shot NAS · NasBench-201 | Kendall's Tau: 0.653 ± 0.012; 0.696 ± 0.016; 0.752 ± 0.018 | No clear effect |
| Use transfer learning to initialize child sub-supernets from parent weights (ablation) | Few-shot NAS with search-based algorithms · NasBench-201 | current best accuracy | No clear effect |
| Replace one supernet with 7 supernets (ablation) | DARTS · CIFAR-10 | test error: 2.76±0.09 -> 2.31±0.08 | Helped |
| Replace one supernet with 7 supernets (ablation) | REA · CIFAR-10 | test error: 2.02±0.03 -> 1.81±0.05 | Helped |
| Replace one supernet with 7 supernets (ablation) | LaNas · CIFAR-10 | test error: 1.68±0.06 -> 1.58±0.04 | Helped |
| Replace one supernet with 5 supernets (ablation) | OFA · ImageNet | Top 1 Acc(%): 80.0 -> 80.50 | Helped |
| Replace one supernet with 5 supernets (ablation) | ProxylessNAS · ImageNet | Top 1 Acc(%): 75.1 -> 75.91 | Helped |
| Replace one supernet with 3 supernets (ablation) | AutoGAN · AutoGAN | FID Score: 12.42 -> 10.73 | Helped |
| Replace one supernet with 3 supernets (ablation) | AutoGAN · AutoGAN | Inception Score: 8.55 -> 8.63 | Helped |
| Use few-shot rather than one-shot NAS (ablation) | DARTS on PTB · Penn Treebank | test Perplexity: 55.7 -> 54.89 | Helped |
Closest papers by meaning
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.95)
- Generalizing Few-Shot NAS with Gradient Matching (ICLR 2022 · similarity 0.95)
- GreedyNAS: Towards Fast One-Shot NAS With Greedy Supernet (CVPR 2020 · similarity 0.93)
- HEP-NAS: Towards Efficient Few-shot Neural Architecture Search via Hierarchical Edge Partitioning (AAAI 2025 · similarity 0.93)
- Subnet-Aware Dynamic Supernet Training for Neural Architecture Search (CVPR 2025 · similarity 0.91)
- Revisiting Neural Networks for Few-Shot Learning: A Zero-Cost NAS Perspective (ICML 2025 · 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.