Research map / Architectures and efficiency / Neural architecture search
EcoNAS: Finding Proxies for Economical Neural Architecture Search
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 |
|---|---|---|---|
| Replace AmoebaNet's c1r0s0e25 proxy with c4r4s0e35 (ablation) | AmoebaNet evolutionary search · CIFAR-10 | top-1 error (%): 3.34 -> 2.94 | Helped |
| Replace AmoebaNet's c1r0s0e25 proxy with c4r4s0e35 (ablation) | AmoebaNet evolutionary search · CIFAR-10 | GPU days: 3150 -> 12 | Helped |
| Replace NASNet proxy with c3r2s1e60 (ablation) | Evolutionary NAS with hierarchical proxy · CIFAR-10 | top-1 error (%): 3.20 -> 2.85 | Helped |
| Use c4r4s0e60 instead of c3r2s1e60 (ablation) | EcoNAS with hierarchical proxy · CIFAR-10 | top-1 error (%): 2.85 -> 2.62 | Helped |
| Use c4r4s0e60 instead of c3r2s1e60 (ablation) | EcoNAS with hierarchical proxy · CIFAR-10 | Spearman coefficient: 0.79 -> 0.85 | Helped |
| Apply the hierarchical proxy strategy (ablation) | Evolutionary NAS using c4r4s0e proxies · CIFAR-10 | top-1 error (%): 3.34 -> 2.94 | Helped |
| Reduce channels and image resolution with the c4r2s0 proxy (ablation) | DARTS · CIFAR-10 | top-1 error (%): 3.00 -> 2.80; 2.76 -> 2.80 | Mixed |
| Reduce channels and image resolution with the c2r2s0 proxy (ablation) | ProxylessNAS · ImageNet-1k | top-1 error (%): 25.4 -> 23.2; 23.3 -> 23.2 | Mixed |
| Use more training samples and fewer epochs at the same iteration count | Reduced-setting model ranking · CIFAR-10 | Spearman coefficient | Helped |
| Reduce channels rather than relying on smaller input resolution | Reduced-setting model ranking · CIFAR-10 | Spearman coefficient | Helped |
Closest papers by meaning
- ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware (ICLR 2019 · similarity 0.93)
- Zero-Cost Proxies for Lightweight NAS (ICLR 2021 · similarity 0.93)
- ProxyBO: Accelerating Neural Architecture Search via Bayesian Optimization with Zero-Cost Proxies (AAAI 2023 · similarity 0.92)
- EZNAS: Evolving Zero-Cost Proxies For Neural Architecture Scoring (NeurIPS 2022 · similarity 0.92)
- AZ-NAS: Assembling Zero-Cost Proxies for Network Architecture Search (CVPR 2024 · similarity 0.90)
- Training-free Neural Architecture Search through Variance of Knowledge of Deep Network Weights (CVPR 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.