Research map / Adversarial robustness and security / Attacks on text and graphs
A New Defense Against Adversarial Images: Turning a Weakness into a Strength
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 |
|---|---|---|---|
| Combine criteria C1 and C2 | ImageNet Inception-v3 adversarial-image detector · ImageNet | detection rate: 0.013 -> 0.199; 0.031 -> 0.468 | Helped |
| Use attack perceptibility threshold τ=0.03 | Combined detector on ResNet-101 · ImageNet, white-box attack | detection rate: 0.571; 0.418 | Helped |
| Restrict the white-box attack to targeting C1 | Combined detector on ResNet-101 · ImageNet, white-box attack | detection rate: 0.512; 0.304 | Helped |
| Use an untargeted attack | Combined detector on ResNet-101 · ImageNet, white-box attack | detection rate: 0.538; 0.342 | Helped |
Closest papers by meaning
- Defense Against Adversarial Images Using Web-Scale Nearest-Neighbor Search (CVPR 2019 · similarity 0.89)
- Robust Feature-Level Adversaries are Interpretability Tools (NeurIPS 2022 · similarity 0.89)
- PixelDefend: Leveraging Generative Models to Understand and Defend against Adversarial Examples (ICLR 2018 · similarity 0.89)
- Evading Adversarial Example Detection Defenses with Orthogonal Projected Gradient Descent (ICLR 2022 · similarity 0.88)
- Detecting Adversarial Data Using Perturbation Forgery (CVPR 2025 · similarity 0.87)
- Constructing Semantics-Aware Adversarial Examples with a Probabilistic Perspective (NeurIPS 2024 · 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.