Research map / Adversarial robustness and security / Attacks on text and graphs
Modeling Biological Immunity to Adversarial Examples
Check 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 retina processing to the ResNet50 input (ablation) | ResNet50 · ILSVRC2012 val subset, PGD-attacked | top-1 accuracy: 9.5 -> 30.2 | Helped |
| Add sparse coding to the retina model (ablation) | Retina Model Only · ILSVRC2012 val subset, PGD-attacked | top-1 accuracy: 30.2 -> 35.4 | Helped |
| Add top-down feedback to retina and sparse coding (ablation) | Retina & Sparse Coding (SC) · ILSVRC2012 val subset, PGD-attacked | top-1 accuracy: 35.4 -> 35.7 | Helped |
| Add top-down feedback to retina and sparse coding (ablation) | Retina & Sparse Coding (SC) · ILSVRC2012 val subset, PGD-attacked | top-5 accuracy: 67.7 -> 65.4 | Hurt |
| Add sparse coding to the retina model (ablation) | Retina Model Only · ILSVRC2012 val subset, original | top-1 accuracy: 46.8 -> 48.8 | Helped |
| Add top-down feedback to retina and sparse coding (ablation) | Retina & Sparse Coding (SC) · ILSVRC2012 val subset, original | top-1 accuracy: 48.8 -> 48.2 | Hurt |
| Use top-down feedback in the ImageNet model (ablation) | Retina & Sparse Coding (SC) · ILSVRC2012 val subset, PGD-attacked | top-1 accuracy: 35.4 -> 35.7 | No clear effect |
Closest papers by meaning
- Attack to Explain Deep Representation (CVPR 2020 · similarity 0.88)
- Adversarially trained neural representations are already as robust as biological neural representations (ICML 2022 · similarity 0.88)
- Towards Imperceptible and Robust Adversarial Example Attacks Against Neural Networks (AAAI 2018 · similarity 0.87)
- Fundamental Tradeoffs between Invariance and Sensitivity to Adversarial Perturbations (ICML 2020 · similarity 0.87)
- Wide Two-Layer Networks can Learn from Adversarial Perturbations (NeurIPS 2024 · similarity 0.85)
- Theoretical Understanding of Learning from Adversarial Perturbations (ICLR 2024 · similarity 0.85)
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.