Research map / Architectures and efficiency / Neural architecture search
RANK-NOSH: Efficient Predictor-Based Architecture Search via Non-Uniform Successive Halving
Ruochen Wang, Xiangning Chen, Minhao Cheng, Xiaocheng Tang, Cho-Jui Hsieh
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 |
|---|---|---|---|
| use NOSH with ranking instead of arch2vec-BO with early stopping at budget 2,969 (ablation) | arch2vec-BO · NAS-Bench-201 CIFAR-10 val | validation accuracy (%): 90.35 -> 91.56 | Improved |
| use NOSH with ranking instead of arch2vec-BO with early stopping at budget 2,969 (ablation) | arch2vec-BO · NAS-Bench-201 CIFAR-100 val | validation accuracy (%): 71.88 -> 73.44 | Improved |
| use NOSH with ranking instead of arch2vec-BO with early stopping at budget 2,969 (ablation) | arch2vec-BO · NAS-Bench-201 ImageNet16-120 val | validation accuracy (%): 45.10 -> 46.43 | Improved |
| use training-free prior scores alone instead of searching (ablation) | mag prior score · NAS-Bench-201 CIFAR-10 test | test accuracy (%): 94.26 -> 93.35 | Hurt |
| shorten schedule E from (10,50,100,200) to (5,10,25,200) (ablation) | RANK-NOSH · NAS-Bench-201 CIFAR-10 val | validation accuracy (%): 91.60 -> 91.57 | No clear effect |
| shorten schedule E from (10,50,100,200) to (5,25,50,200) (ablation) | RANK-NOSH · NAS-Bench-201 CIFAR-10 val | validation accuracy (%): 91.60 -> 91.59 | No clear effect |
| drop the 100-epoch level, E=(10,50,200) (ablation) | RANK-NOSH · NAS-Bench-201 CIFAR-10 val | validation accuracy (%): 91.60 -> 91.60 | No clear effect |
| lower move ratio r from 0.5 to 0.3 (ablation) | RANK-NOSH · NAS-Bench-201 CIFAR-10 val | validation accuracy (%): 91.60 -> 91.40 | Hurt |
| raise move ratio r from 0.5 to 0.7 (ablation) | RANK-NOSH · NAS-Bench-201 CIFAR-10 val | validation accuracy (%): 91.60 -> 91.58 | No clear effect |
| use RANK-NOSH instead of arch2vec-BO at lower budget | arch2vec-BO · NAS-Bench-201 CIFAR-100 test | test accuracy (%): 73.41 -> 73.51 | Improved |
| use RANK-NOSH instead of arch2vec-BO on DARTS space | arch2vec-BO · DARTS space CIFAR-10 avg | test error (%): 2.56 -> 2.53 | Improved |
| use RANK-NOSH instead of arch2vec-BO on NAS-Bench-101 | arch2vec-BO · NAS-Bench-101 | test accuracy (%): 94.05 -> 93.97 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- BANANAS — BANANAS: Bayesian Optimization with Neural Architectures for Neural Architecture Search (AAAI 2021) · better on 1 of 2 shared results
- DARTS — DARTS-: Robustly Stepping out of Performance Collapse Without Indicators (ICLR 2021) · better on 1 of 1 shared result
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
- How Powerful are Performance Predictors in Neural Architecture Search? (NeurIPS 2021 · similarity 0.90)
- A Generic Graph-Based Neural Architecture Encoding Scheme With Multifaceted Information (TPAMI 2022 · similarity 0.90)
- Per-Architecture Training-Free Metric Optimization for Neural Architecture Search (NeurIPS 2025 · similarity 0.89)
- Encodings for Prediction-based Neural Architecture Search (ICML 2024 · similarity 0.88)
- Boosting Order-Preserving and Transferability for Neural Architecture Search: a Joint Architecture Refined Search and Fine-tuning Approach (CVPR 2024 · similarity 0.88)
- Robustifying and Boosting Training-Free Neural Architecture Search (ICLR 2024 · similarity 0.88)
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.
