Research map / Learning theory and algorithms / Kernels and regression theory
Generalization Guarantees for Neural Architecture Search with Train-Validation Split
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 |
|---|---|---|---|
| Increase DARTS validation sample size | DARTS · CIFAR-10 | test-validation error gap: 20 to 1000; around n_\mathcal{V}=250 | Helped |
| Use validation sizes at or below 100 | DARTS · CIFAR-10 | test risk: n_\mathcal{V}\leq 100 | Hurt |
| Train for more epochs | DARTS · CIFAR-10 | test-validation error gap | Hurt |
| Use a spectral estimator on validation data to select the hyperparameter, then solve ERM on training data | two-stage rank-1 learning · Overparameterized rank-1 learning | population risk: 1-\frac{1}{\bar{p}} -> 1-\frac{1}{\bar{p}}+\frac{200\bar{h}}{1-1/\bar{p}} | Helped |
| Increase the product of feature dimensions and hyperparameter dimensions beyond the stated regime | two-stage rank-1 learning · Overparameterized rank-1 learning | hyperparameter recovery correlation: ph\lesssim{n^2} | Hurt |
Closest papers by meaning
- Generalization Properties of NAS under Activation and Skip Connection Search (NeurIPS 2022 · similarity 0.91)
- Speedy Performance Estimation for Neural Architecture Search (NeurIPS 2021 · similarity 0.91)
- MiLeNAS: Efficient Neural Architecture Search via Mixed-Level Reformulation (CVPR 2020 · similarity 0.91)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.89)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.89)
- Robustifying and Boosting Training-Free Neural Architecture Search (ICLR 2024 · similarity 0.89)
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.