Research map / Adversarial robustness and security / Backdoor and poisoning
FRBAT: Conditionally-Visible Physical Backdoor Attack via Fluorescence
Yalun Wu, Liu Liu, Endong Tong, Yingxiao Xiang, Xiaoting Lyu, Zhen Han, Jiqiang Liu
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 |
|---|---|---|---|
| raise poisoning rate from 1% to 5% (ablation) | FRBAT · BelgiumTS (ResNet-18) | ASR: 8.82 -> 95.29 | Improved |
| raise poisoning rate from 1% to 5% (ablation) | FRBAT · GTSRB (ResNet-18) | ASR: 91.08 -> 94.52 | Improved |
| raise poisoning rate (ablation) | FRBAT · GTSRB (ResNet-18) | clean ACC: 91.64 -> 93.13 | No clear effect |
| enlarge trigger area ratio from 0.01 to 0.05 (ablation) | FRBAT · BelgiumTS (ResNet-18) | ASR: 5.61 -> 95.29 | Improved |
| enlarge trigger area ratio from 0.01 to 0.05 (ablation) | FRBAT · CTSRD (ResNet-18) | ASR: 95.00 -> 97.47 | No clear effect |
| apply I-BAU defense | FRBAT · GTSRB (ResNet-18) | ASR: 94.52 -> 12.07 | Hurt |
| apply FT-SAM defense | FRBAT · BelgiumTS (ResNet-18) | ASR: 95.29 -> 93.86 | No clear effect |
| apply MCR and NAD defenses | FRBAT · CTSRD (ResNet-18) | ASR: 97.47 -> 98.88 | No clear effect |
| use fluorescence trigger vs FTrojan | FRBAT · GTSRB (digital avg) | ASR: 50.26 -> 97.08 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- COMBAT — COMBAT: Alternated Training for Effective Clean-Label Backdoor Attacks (AAAI 2024) · better on 3 of 3 shared results
- BppAttack — BppAttack: Stealthy and Efficient Trojan Attacks Against Deep Neural Networks via Image Quantization and Contrastive Adversarial Learning (CVPR 2022) · better on 2 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- An Invisible Black-Box Backdoor Attack through Frequency Domain (ECCV 2022 · similarity 0.87)
- A Dual Stealthy Backdoor: From Both Spatial and Frequency Perspectives (AAAI 2024 · similarity 0.87)
- Conditional Backdoor Attack via JPEG Compression (AAAI 2024 · similarity 0.86)
- The Fluorescent Veil: A Stealthy and Effective Physical Adversarial Patch Against Traffic Sign Recognition (NeurIPS 2025 · similarity 0.85)
- Not All Prompts Are Secure: A Switchable Backdoor Attack Against Pre-trained Vision Transfomers (CVPR 2024 · similarity 0.84)
- UIBDiffusion: Universal Imperceptible Backdoor Attack for Diffusion Models (CVPR 2025 · similarity 0.84)
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.
