Research map / Adversarial robustness and security / Certified and empirical robustness
Neural Architecture Design and Robustness: A Dataset
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 |
|---|---|---|---|
| Search architectures using FGSM ε=1.0 accuracy instead of clean accuracy | NAS-Bench-201 search (Local Search) · CIFAR-10 test, FGSM ε=1.0 | accuracy: 63.95 -> 69.10 | Helped |
| Search architectures using FGSM ε=1.0 accuracy instead of clean accuracy | NAS-Bench-201 search · CIFAR-10 test, PGD ε=1.0 | accuracy: 40.09–41.17 -> 45.59–48.27 | Helped |
| Search architectures using FGSM ε=1.0 accuracy instead of clean accuracy | NAS-Bench-201 search · CIFAR-10 test, APGD ε=1.0 | accuracy: 17.84–18.74 -> 20.72–23.18 | Helped |
| Search architectures using FGSM ε=1.0 accuracy instead of clean accuracy | NAS-Bench-201 search · CIFAR-10 test, clean | accuracy: 94.21 -> 93.52 | Hurt |
| Replace the edge 3 3×3 convolution with average pooling (ablation) | CIFAR-10 clean-best architecture #13714 · CIFAR-10 adversarial attacks | robust accuracy | Helped |
| Make a single neighboring architecture change to clean-best architecture #13714 (ablation) | CIFAR-10 clean-best architecture #13714 · CIFAR-10-C | accuracy | Hurt |
| Use carefully chosen topology while holding cell kernel parameter count at 18 (ablation) | NAS-Bench-201 architectures · CIFAR-10 | mean adversarial accuracy: [0.21,0.4] | Mixed |
Closest papers by meaning
- When NAS Meets Robustness: In Search of Robust Architectures Against Adversarial Attacks (CVPR 2020 · similarity 0.92)
- Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks (NeurIPS 2021 · similarity 0.92)
- Revisiting Residual Networks for Adversarial Robustness (CVPR 2023 · similarity 0.92)
- Robust NAS under adversarial training: benchmark, theory, and beyond (ICLR 2024 · similarity 0.91)
- DSRNA: Differentiable Search of Robust Neural Architectures (CVPR 2021 · similarity 0.91)
- Towards Accurate and Robust Architectures via Neural Architecture Search (CVPR 2024 · 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.