Research map / Architectures and efficiency / Neural architecture search
Extensible and Efficient Proxy for Neural Architecture Search
Yuhong Li, Jiajie Li, Cong Hao, Pan Li, Jinjun Xiong, Deming Chen
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 |
|---|---|---|---|
| add untrained barrier layer to few-shot regression proxy (ablation) | Eproxy · NAS-Bench-101, 10 iterations, LR 1 | Spearman rho: 0.08 -> 0.65 | Improved |
| add Discrete Proxy Search to Eproxy (ablation) | Eproxy · NAS-Bench-101 | Spearman rho: 0.65 -> 0.69 | Improved |
| add Discrete Proxy Search to Eproxy (ablation) | Eproxy · NAS-Bench-Trans-Micro average | Spearman rho: 0.29 -> 0.69 | Improved |
| add Discrete Proxy Search to Eproxy (ablation) | Eproxy · NAS-Bench-MR average | Spearman rho: 0.27 -> 0.51 | Improved |
| add Discrete Proxy Search to Eproxy (ablation) | Eproxy · NDS ImageNet average (transferred) | Spearman rho: 0.55 -> 0.73 | Improved |
| Eproxy without DPS versus NASWOT on NDS CIFAR-10 | Eproxy · NDS CIFAR-10 average | Spearman rho: 0.43 -> 0.41 | No clear effect |
| Eproxy without DPS on NAS-Bench-Trans-Micro versus other proxies in ranking correlation | Eproxy · NAS-Bench-Trans-Micro average | Spearman rho: 0.47 -> 0.29 | Hurt |
| use Eproxy+DPS instead of early stopping on target tasks | Eproxy+DPS · NAS-Bench-Trans-Micro average | Spearman rho: 0.43 -> 0.69 | Improved |
| use Eproxy+DPS as fitness for regularized evolution on NAS-Bench-201 | Eproxy+DPS · NAS-Bench-201 CIFAR-10 | queries to optimum: 563.2 -> 58.0 + 20 | Improved |
| NAS with Eproxy+DPS on DARTS-ImageNet | Eproxy+DPS · DARTS ImageNet top-1 error | test error: 25.7 -> 24.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ZiCo — ZiCo: Zero-shot NAS via inverse Coefficient of Variation on Gradients (ICLR 2023) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- EcoNAS: Finding Proxies for Economical Neural Architecture Search (CVPR 2020 · similarity 0.91)
- ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware (ICLR 2019 · similarity 0.91)
- Revolutionizing Training-Free NAS: Towards Efficient Automatic Proxy Discovery via Large Language Models (NeurIPS 2025 · similarity 0.90)
- Training-free Neural Architecture Search through Variance of Knowledge of Deep Network Weights (CVPR 2025 · similarity 0.87)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.87)
- einspace: Searching for Neural Architectures from Fundamental Operations (NeurIPS 2024 · similarity 0.87)
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.
