Research map / Adversarial robustness and security / Certified and empirical robustness
IrisFP: Adversarial-Example-based Model Fingerprinting with Enhanced Uniqueness and Robustness
Ziye Geng, Guang Yang, Yihang Chen, Changqing Luo
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 |
|---|---|---|---|
| add fingerprint selection to seed fingerprints (ablation) | IrisFP seeds · CIFAR-10 (ResNet-18) | AUC: 0.691 -> 0.748 | Improved |
| replace the global threshold with fingerprint-specific thresholds (ablation) | IrisFP · CIFAR-10 (ResNet-18) | AUC: 0.812 -> 0.893 | Improved |
| extend seeds into composite-sample fingerprints (ablation) | IrisFP · five datasets (ResNet-18) | AUC | Improved |
| add fingerprint selection to composite-sample fingerprints (ablation) | IrisFP · five datasets (ResNet-18) | AUC | Improved |
| raise K from 10 to 20 while lowering T from 20 to 10 (200 queries) | IrisFP · CIFAR-100 (ResNet-18) | AUC: 0.842 -> 0.905 | Improved |
| use K=40, T=5 instead of K=20, T=10 | IrisFP · CIFAR-100 (ResNet-18) | AUC: 0.905 -> 0.916 | Improved |
| use K=100, T=2 instead of K=40, T=5 | IrisFP · CIFAR-100 (ResNet-18) | AUC: 0.916 -> 0.872 | Hurt |
| increase queries from 20 to 200 | IrisFP · Fashion-MNIST (MobileNet-V2) | AUC: 0.838 -> 0.963 | Improved |
| increase queries from 200 to 800 | IrisFP · Fashion-MNIST (MobileNet-V2) | AUC: less than 0.8% | No clear effect |
| enlarge reference model sets from 1 to 12 models per set | IrisFP · CIFAR-100 (ResNet-18) | AUC: 0.792 -> 0.918 | Improved |
| enlarge reference model sets from 1 to 12 models per set | IrisFP · Fashion-MNIST (MobileNet-V2) | AUC: 0.830 -> 0.969 | Improved |
| use IrisFP instead of best baseline (multi-boundary composite fingerprints) | fingerprinting · CIFAR-100 (ResNet-18) | AUC: 0.916; 13.7% over best baseline | Improved |
Closest papers by meaning
- Fingerprinting Deep Neural Networks for Ownership Protection: An Analytical Approach (ICLR 2026 · similarity 0.91)
- UniReFP: Robust Unified Fingerprinting for Vision Models against Cross-Task Repurposing Attacks (NeurIPS 2026 · similarity 0.87)
- Intersecting-Boundary-Sensitive Fingerprinting for Tampering Detection of DNN Models (ICML 2024 · similarity 0.85)
- Scalable Fingerprinting of Large Language Models (NeurIPS 2025 · similarity 0.85)
- SDBF: Steep-Decision-Boundary Fingerprinting for Hard-Label Tampering Detection of DNN Models (CVPR 2025 · similarity 0.85)
- SIF: Semantically In-Distribution Fingerprints for Large Vision-Language Models (CVPR 2026 · 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.
