Research map / Architectures and efficiency / Neural architecture search
Neural Architecture Search in A Proxy Validation Loss Landscape
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 |
|---|---|---|---|
| add second-order validation loss (1-epoch training of sampled architecture) to estimation (ablation) | PVLL-NAS search · CIFAR-10 | test error (%): 3.48 -> 2.72 | Helped |
| add warm-up stage (100 random architectures) before search (ablation) | PVLL-NAS sampler · CIFAR-10 | test error (%): 3.10 -> 2.72 | Helped |
| add weighted MSE loss (weight 1/p_t) (ablation) | PVLL-NAS estimator · CIFAR-10 | test error (%): 2.81 -> 2.72 | Helped |
| replace proposed sampler with uniform sampling (ablation) | PVLL-NAS · CIFAR-10 | test error (%): 2.72 -> 3.08 | Hurt |
| increase warm-up population size from 100 to 500 (ablation) | validation loss estimator · population size sweep | estimation loss: 0.0009 -> 0.0007 at 5x GPU hours (1.2 to 11.6 hours range) | No clear effect |
| search with proxy validation loss landscape vs. DARTS/GDAS | NAS on CIFAR-10 and ImageNet · CIFAR-10 and ImageNet 2012 | test error: 2.70 vs GDAS+cutout 2.93 on CIFAR-10; 25.6 vs DARTS 26.7 top-1 on ImageNet | Helped |
Closest papers by meaning
- Generalization Guarantees for Neural Architecture Search with Train-Validation Split (ICML 2021 · similarity 0.90)
- Speedy Performance Estimation for Neural Architecture Search (NeurIPS 2021 · similarity 0.90)
- EcoNAS: Finding Proxies for Economical Neural Architecture Search (CVPR 2020 · similarity 0.89)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.87)
- Encodings for Prediction-based Neural Architecture Search (ICML 2024 · similarity 0.87)
- NEAR: A Training-Free Pre-Estimator of Machine Learning Model Performance (ICLR 2025 · 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.