Research map / Adversarial robustness and security / Certified and empirical robustness
CIFS: Improving Adversarial Robustness of CNNs via Channel-wise Importance-based Feature Selection
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 |
|---|---|---|---|
| Add CIFS channel selection to adversarially trained ResNet-18 | ResNet-18 · CIFAR10 | PGD-100 robust accuracy: 44.72 -> 48.70 | Helped |
| Add CIFS channel selection to adversarially trained WRN-28-10 | WRN-28-10 · CIFAR10 | PGD-100 robust accuracy: 47.08 -> 51.51 | Helped |
| Add CIFS channel selection to adversarially trained ResNet-18 | ResNet-18 · SVHN | PGD-100 robust accuracy: 45.81 -> 47.49 | Helped |
| Add CIFS to FAT training | ResNet-18 (FAT) · CIFAR10 | PGD-100 robust accuracy: 45.35 -> 49.52 | Helped |
| Replace no channel mask with sigmoid importance mask (ablation) | ResNet-18 · CIFAR10 | PGD-20 robust accuracy: 46.64 -> 49.87 | Helped |
| Use softplus importance mask instead of sigmoid (ablation) | CIFS ResNet-18 · CIFAR10 | PGD-20 robust accuracy: 49.87 -> 50.38 | Helped |
| Use softmax importance mask instead of sigmoid (ablation) | CIFS ResNet-18 · CIFAR10 | PGD-20 robust accuracy: 49.87 -> 51.23 | Helped |
| Use top-2 rather than top-1 relevance assessment (ablation) | CIFS ResNet-18 · CIFAR10 | PGD-20 robust accuracy: 47.24 -> 51.23 | Helped |
| Use top-3 rather than top-2 relevance assessment (ablation) | CIFS ResNet-18 · CIFAR10 | PGD-20 robust accuracy: 51.23 -> 49.91 | Hurt |
| Set beta to 2 rather than 0.1 (ablation) | CIFS ResNet-18 · CIFAR10 | PGD-100 robust accuracy: 46.28 -> 48.74 | Helped |
| Set beta to 100 rather than 2 (ablation) | CIFS ResNet-18 · CIFAR10 | PGD-100 robust accuracy: 48.74 -> 43.12 | Hurt |
| Apply CIFS at P1 and P2 rather than only P1 (ablation) | ResNet-18 · CIFAR10 | PGD-100 robust accuracy: 45.95 -> 48.74 | Helped |
Closest papers by meaning
- Robust Models are less Over-Confident (NeurIPS 2022 · similarity 0.90)
- A Combination of Noise and Bilateral Filters Achieve Supralinear and Scalable Adversarial Robustness in CNNs (CVPR 2026 · similarity 0.90)
- Robust NAS under adversarial training: benchmark, theory, and beyond (ICLR 2024 · similarity 0.88)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.88)
- Learning Better Certified Models from Empirically-Robust Teachers (NeurIPS 2026 · similarity 0.88)
- Boosting Adversarial Robustness with CLAT: Criticality Leveraged Adversarial Training (ICML 2025 · similarity 0.88)
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.