Research map / Architectures and efficiency / Neural architecture search
Geometry-Aware Gradient Algorithms for 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 PC-DARTS architecture updates with simplex-constrained exponentiated-gradient updates | PC-DARTS · CIFAR-10, DARTS search space | test error (%): 2.57±0.07 -> 2.50±0.06 | Helped |
| Replace PC-DARTS architecture updates with simplex-constrained exponentiated-gradient updates | PC-DARTS · ImageNet, DARTS search space, direct search | top-1 test error (%): 24.2 -> 24.0 | Helped |
| Replace PC-DARTS architecture updates with simplex-constrained exponentiated-gradient updates | PC-DARTS · ImageNet, DARTS search space, CIFAR-10 search transfer | top-1 test error (%): 25.1 -> 24.3 | Helped |
| Apply GAEA to DARTS with single-level ERM instead of the baseline optimization | DARTS (ERM) · NAS-Bench-201 CIFAR-10 test, transfer search | accuracy (%): 84.39±3.82 -> 94.10±0.29 | Helped |
| Apply GAEA to DARTS with single-level ERM instead of the baseline optimization | DARTS (ERM) · NAS-Bench-201 CIFAR-100 test, transfer search | accuracy (%): 54.81±7.08 -> 72.60±0.89 | Helped |
| Apply GAEA to DARTS with single-level ERM instead of the baseline optimization | DARTS (ERM) · NAS-Bench-201 ImageNet-16-120 test, transfer search | accuracy (%): 31.82±4.78 -> 45.81±0.51 | Helped |
| Apply GAEA to DARTS with single-level ERM instead of the baseline optimization | DARTS (ERM) · NAS-Bench-201 CIFAR-100 test, direct search | accuracy (%): 51.26±6.14 -> 73.43±0.13 | Helped |
| Apply GAEA to DARTS with single-level ERM instead of the baseline optimization | DARTS (ERM) · NAS-Bench-201 ImageNet-16-120 test, direct search | accuracy (%): 31.35±7.46 -> 46.36±0.00 | Helped |
| Apply GAEA to GDAS | GDAS · NAS-Bench-201 CIFAR-100 test, transfer search | accuracy (%): 67.52±0.15 -> 70.47±0.47 | Helped |
| Apply GAEA to GDAS | GDAS · NAS-Bench-201 CIFAR-10 test, transfer search | accuracy (%): 93.61±0.09 -> 93.55±0.13 | No clear effect |
| Apply GAEA to GDAS | GDAS · NAS-Bench-201 ImageNet-16-120 test, transfer search | accuracy (%): 40.91±0.12 -> 40.91±0.12 | No clear effect |
| Use single-level ERM rather than bilevel optimization (ablation) | DARTS · NAS-Bench-201 CIFAR-10 test, transfer search | accuracy (%): 54.30±0.00 -> 84.39±3.82 | Helped |
Closest papers by meaning
- Adaptive Stochastic Natural Gradient Method for One-Shot Neural Architecture Search (ICML 2019 · similarity 0.91)
- Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective (ICLR 2021 · similarity 0.91)
- Vision-Oriented Lightweight Neural Architecture Search with Budget-Adaptive Evaluation (CVPR 2026 · similarity 0.91)
- Posterior-Guided Neural Architecture Search (AAAI 2020 · similarity 0.90)
- MiLeNAS: Efficient Neural Architecture Search via Mixed-Level Reformulation (CVPR 2020 · similarity 0.90)
- TransNAS-Bench-101: Improving Transferability and Generalizability of Cross-Task Neural Architecture Search (CVPR 2021 · 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.