Research map / Adversarial robustness and security / Attacks on text and graphs
Spatially Transformed Adversarial Examples
Chaowei Xiao, Jun-Yan Zhu, Bo Li, Warren He, Mingyan Liu, Dawn Song
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 |
|---|---|---|---|
| Use smooth spatial-flow perturbations instead of direct pixel-value perturbations | Targeted adversarial attacks on MNIST and CIFAR-10 · MNIST and CIFAR-10 | attack success rate under adversarial-training defenses | Improved |
| Use smooth spatial-flow perturbations instead of direct pixel-value perturbations | Adversarial examples evaluated by human participants · ImageNet-compatible dataset | human realism choice rate: 47.01\% \pm 1.96\% | No clear effect |
| Use spatial-flow attacks rather than FGSM or C&W | Models with adversarial-training defenses · MNIST and CIFAR-10 | attack success rate | Improved |
| Apply a 3 × 3 average-filter restoration to stAdv examples | stAdv adversarial examples · MNIST and CIFAR-10 | accuracy on recovered images | Mixed |
Closest papers by meaning
- Adversarial Attacks Beyond the Image Space (CVPR 2019 · similarity 0.90)
- Leveraging Spatial Invariance to Boost Adversarial Transferability (ICCV 2025 · similarity 0.89)
- Characterizing Adversarial Examples Based on Spatial Consistency Information for Semantic Segmentation (ECCV 2018 · similarity 0.88)
- Detecting Adversarial Examples Through Image Transformation (AAAI 2018 · similarity 0.88)
- Pixel2Feature Attack (P2FA): Rethinking the Perturbed Space to Enhance Adversarial Transferability (ICML 2025 · similarity 0.86)
- Breaking Barriers in Physical-World Adversarial Examples: Improving Robustness and Transferability via Robust Feature (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.
