Research map / Adversarial robustness and security / Certified and empirical robustness
Architectural Adversarial Robustness: The Case for Deep Pursuit
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 |
|---|---|---|---|
| Use deep pursuit with residual connections instead of layered basis pursuit | CIFAR-10 networks under FGSM attacks · CIFAR-10 | adversarial accuracy | Helped |
| Add residual connections to deep pursuit (ablation) | Deep pursuit · CIFAR-10 | adversarial accuracy | Helped |
| Add residual connections to feed-forward networks (ablation) | Feed-forward networks · CIFAR-10 | adversarial accuracy | Helped |
| Increase network depth (ablation) | Residual feed-forward networks · CIFAR-10 | adversarial accuracy | Helped |
| Increase global optimization iterations (ablation) | Deep pursuit · CIFAR-10 | adversarial accuracy | Helped |
| Use residual connections in deep pursuit (ablation) | Deep pursuit · CIFAR-10 | iterations needed to outperform layered basis pursuit | Helped |
| Remove residual connections from deep pursuit (ablation) | Deep pursuit · CIFAR-10 | adversarial accuracy across iterations | Hurt |
| Use deep pursuit instead of layered basis pursuit | CIFAR-10 networks under FGSM attacks · CIFAR-10 | reconstruction error | Helped |
| Add skip connections to deep pursuit (ablation) | Deep pursuit · CIFAR-10 | reconstruction error | Helped |
| Optimize the global dictionary synchronously rather than layer by layer | Deep pursuit versus layered basis pursuit · CIFAR-10 | deep frame potential | Helped |
| Add skip connections (ablation) | Deep pursuit · CIFAR-10 | deep mutual coherence | Helped |
| Continue iterative optimization after training and validation error have converged (ablation) | Deep pursuit · CIFAR-10 | adversarial accuracy | Helped |
Closest papers by meaning
- Revisiting Residual Networks for Adversarial Robustness (CVPR 2023 · similarity 0.90)
- Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks (NeurIPS 2021 · similarity 0.90)
- Robust NAS under adversarial training: benchmark, theory, and beyond (ICLR 2024 · similarity 0.88)
- Towards Accurate and Robust Architectures via Neural Architecture Search (CVPR 2024 · similarity 0.88)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.87)
- First Line of Defense: A Robust First Layer Mitigates Adversarial Attacks (AAAI 2025 · similarity 0.87)
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.