Research map / Learning theory and algorithms / Kernels and regression theory
Generalization Properties of NAS under Activation and Skip Connection 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 |
|---|---|---|---|
| Guide architecture selection with the proposed minimum-eigenvalue estimate and select k=20 candidates | Train-free NAS on NAS-Bench-201 · CIFAR-100 | accuracy (%): 70.78 -> 71.42±0.63 | Helped |
| Guide architecture selection with the proposed minimum-eigenvalue estimate and select k=20 candidates | Train-free NAS on NAS-Bench-201 · ImageNet-16 | accuracy (%): 44.63 -> 45.54±0.04 | Helped |
| Guide architecture selection with the proposed minimum-eigenvalue estimate and select k=20 candidates | Train-free NAS on NAS-Bench-201 · CIFAR-10 | accuracy (%): 93.38 -> 93.46±0.01 | Helped |
| Select k=5 rather than k=20 candidate architectures | Eigen-NAS · CIFAR-100 | accuracy (%): 71.42±0.63 -> 69.92±1.82 | Hurt |
| Select k=5 rather than k=20 candidate architectures | Eigen-NAS · ImageNet-16 | accuracy (%): 45.54±0.04 -> 45.53±0.06 | No clear effect |
| Select ReLU and LeakyReLU as activation functions | DARTS activation search · Fashion-MNIST | activation selection probability | Helped |
| Use skip connections in each layer | DARTS skip-connection search · Fashion-MNIST | skip-connection selection | Helped |
Closest papers by meaning
- Generalization Guarantees for Neural Architecture Search with Train-Validation Split (ICML 2021 · similarity 0.91)
- Unifying and Boosting Gradient-Based Training-Free Neural Architecture Search (NeurIPS 2022 · similarity 0.90)
- Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective (ICLR 2021 · similarity 0.90)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.90)
- NAS-Bench-201: Extending the Scope of Reproducible Neural Architecture Search (ICLR 2020 · similarity 0.89)
- Neural Architecture Search without Training (ICML 2021 · 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.