Research map / Adversarial robustness and security / Certified and empirical robustness
Gabor Layers Enhance Network Robustness
Juan C. Pérez, Motasem Alfarra, Guillaume Jeanneret, Adel Bibi, Ali Thabet, Bernard Ghanem, Pablo Arbeláez
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| replace first conv layers with Gabor layers | VGG16 · CIFAR10 PGD eps=8/255 | adversarial accuracy: 23.63 -> 30.11 | Improved |
| replace first conv layers with Gabor layers | VGG16 · SVHN PGD eps=8/255 | adversarial accuracy: 5.84 -> 14.57 | Improved |
| replace first conv layer with Gabor layer | LeNet · MNIST PGD eps=0.2 | adversarial accuracy: 4.39 -> 7.94 | Improved |
| replace first conv layer with Gabor layer | WideResNet · SVHN PGD eps=8/255 | adversarial accuracy: 1.02 -> 1.03 | No clear effect |
| replace first conv layer with Gabor layer | AlexNet · CIFAR100 PGD eps=2/255 | adversarial accuracy: 15.05 -> 14.77 | Hurt |
| replace conv layers with Gabor layers | VGG16 · CIFAR100 clean test | accuracy: 67.54 -> 64.49 | Hurt |
| replace conv layer with Gabor layer | LeNet · MNIST clean test | accuracy: 99.36 -> 99.03 | Hurt |
| add Lipschitz regularizer to Gabor-layered network | LeNet · MNIST PGD eps=0.2 | adversarial accuracy: 7.94 -> 22.69 | Improved |
| add Lipschitz regularizer to Gabor-layered network | VGG16 · SVHN PGD eps=8/255 | adversarial accuracy: 14.57 -> 15.99 | Improved |
| add Lipschitz regularizer to Gabor-layered network | AlexNet · CIFAR100 PGD eps=2/255 | adversarial accuracy: 14.77 -> 14.68 | No clear effect |
| add Gabor layers to adversarially trained network | AlexNet · CIFAR10 adv. trained | adversarial accuracy: 19.24 -> 20.26 | Improved |
| add Gabor layers to adversarially trained network | AlexNet · CIFAR100 adv. trained | adversarial accuracy: 13.48 -> 10.94 | Hurt |
Closest papers by meaning
- A Combination of Noise and Bilateral Filters Achieve Supralinear and Scalable Adversarial Robustness in CNNs (CVPR 2026 · similarity 0.90)
- Revisiting Residual Networks for Adversarial Robustness (CVPR 2023 · similarity 0.89)
- CIFS: Improving Adversarial Robustness of CNNs via Channel-wise Importance-based Feature Selection (ICML 2021 · similarity 0.88)
- Neural Architecture Design and Robustness: A Dataset (ICLR 2023 · similarity 0.87)
- Robust NAS under adversarial training: benchmark, theory, and beyond (ICLR 2024 · similarity 0.86)
- First Line of Defense: A Robust First Layer Mitigates Adversarial Attacks (AAAI 2025 · similarity 0.86)
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.
