Research map / Adversarial robustness and security / Certified and empirical robustness
A Game Theoretic Analysis of Additive Adversarial Attacks and Defenses
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 |
|---|---|---|---|
| Apply SMOOTH against FGM | 4-layer CNN · MNIST binary class pairs | Approximate accuracy (%): 53.3 (10.0) -> 71.2 (14.2) | Helped |
| Apply SMOOTH against FGM | 4-layer CNN · FMNIST binary class pairs | Approximate accuracy (%): 47.4 (5.1) -> 67.4 (9.0) | Helped |
| Apply SMOOTH against PGD | 4-layer CNN · MNIST binary class pairs | Approximate accuracy (%): 71.9 (12.0) -> 94.0 (4.0) | Helped |
| Apply SMOOTH against PGD | 4-layer CNN · FMNIST binary class pairs | Approximate accuracy (%): 74.7 (7.3) -> 90.3 (8.5) | Helped |
| Apply SMOOTH against FGM | 4-layer CNN · MNIST 0 vs 1 | True accuracy (%): 63.0 -> 95.6 | Helped |
| Apply SMOOTH against PGD | 4-layer CNN · MNIST 0 vs 1 | True accuracy (%): 47.7 -> 75.1 | Helped |
| Apply SMOOTH against FGM | 4-layer CNN · MNIST 0 vs 1 | Approximate accuracy (%): 48.3 -> 94.5 | Helped |
| Apply SMOOTH against PGD | 4-layer CNN · MNIST 0 vs 1 | Approximate accuracy (%): 85.6 -> 99.1 | Helped |
| Use PGD instead of FGM against the undefended classifier | 4-layer CNN · MNIST 0 vs 1 | True accuracy (%): 63.0 -> 47.7 | Hurt |
| Use PGD instead of FGM against the undefended classifier | 4-layer CNN · MNIST 0 vs 1 | Approximate accuracy (%): 48.3 -> 85.6 | Mixed |
Closest papers by meaning
- Mixed Nash Equilibria in the Adversarial Examples Game (ICML 2021 · similarity 0.92)
- Adversaries With Incentives: A Strategic Alternative to Adversarial Robustness (ICLR 2025 · similarity 0.87)
- Quickest Detection of Adversarial Attacks Against Correlated Equilibria (AAAI 2025 · similarity 0.86)
- Rethinking Adversarial Policies: A Generalized Attack Formulation and Provable Defense in RL (ICLR 2024 · similarity 0.86)
- Practical Adversarial Attacks on Stochastic Bandits via Fake Data Injection (NeurIPS 2026 · similarity 0.86)
- Optimal Attack and Defense for Reinforcement Learning (AAAI 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.