Research map / Architectures and efficiency / Neural architecture search
Neural Architecture Search as Sparse Supernet
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 |
|---|---|---|---|
| Use AdamHAPG rather than HAPG for architecture optimization | SparseNAS · CIFAR-10 | test error (%): 2.73 ± 0.05 -> 2.69 ± 0.03 | Helped |
| Search a cherry-picked search space | SparseNAS + AdamHAPG · CIFAR-10 | test error (%): 2.69 ± 0.03 -> 2.50 | Helped |
| Use AdamHAPG instead of Adam (ablation) | SparseNAS · CIFAR-10, α=0.3 | test error (%): 3.32 -> 3.30 | No clear effect |
| Use AdamHAPG instead of Adam (ablation) | SparseNAS · CIFAR-10, α=0.5 | test error (%): 2.93 -> 2.69 | Helped |
| Use AdamHAPG instead of Adam (ablation) | SparseNAS · CIFAR-10, α=0.7 | test error (%): 3.93 -> 2.79 | Helped |
| Set α to 0.5 rather than 0.3 or 0.7 (ablation) | SparseNAS + AdamHAPG · CIFAR-10 | test error (%): 3.30, 2.69, 2.79 | Helped |
| Use a larger λ increase step of 0.03 (ablation) | SparseNAS search · CNN cell search | valid accuracy stability | Hurt |
| Use smaller λ increase steps of 0.01 or 0.02 (ablation) | SparseNAS search · CNN cell search | valid accuracy stability | Helped |
| Use Sparse Group Lasso instead of Lasso or Group Lasso (ablation) | HAPG sparse network optimization · DIGITS | element and group sparsity | Helped |
| Use HAPG or AdamHAPG instead of SGD or Adam (ablation) | SGL-constrained network pruning · DIGITS | structure compactness | Helped |
| Use AdamHAPG rather than HAPG for recurrent architecture search (ablation) | SparseNAS · PTB | search training stability | Helped |
Closest papers by meaning
- ISTA-NAS: Efficient and Consistent Neural Architecture Search by Sparse Coding (NeurIPS 2020 · similarity 0.92)
- Boosting Order-Preserving and Transferability for Neural Architecture Search: a Joint Architecture Refined Search and Fine-tuning Approach (CVPR 2024 · similarity 0.91)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.90)
- Subnet-Aware Dynamic Supernet Training for Neural Architecture Search (CVPR 2025 · similarity 0.90)
- Efficient Neural Architecture Search via Proximal Iterations (AAAI 2020 · similarity 0.90)
- SUMNAS: Supernet with Unbiased Meta-Features for Neural Architecture Search (ICLR 2022 · 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.