Research map / Federated learning and privacy / Privacy attacks and secure learning
Proteus: Model Leakage-Induced Adversarial Attack in Federated Learning
Junjie Shan, Yue Zhang, Ziqi Zhao, Ka Ho Chow
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 Proteus instead of PGD on leaked 30% model | leaked FL model · RAF-DB | transfer ASR (%): 59.75 -> 85.40 | Improved |
| use leaked intermediate model instead of locally trained surrogate | surrogate model · RAF-DB | transfer ASR (%): 8.35 -> 59.75 | Improved |
| add Phase 1 persistent direction discovery (transform ensemble) (ablation) | Proteus · RAF-DB | transfer ASR (%): 75.82 -> 85.40 | Improved |
| add Phase 2 adversarial margin expansion (ablation) | Proteus · RAF-DB | transfer ASR (%): 82.23 -> 85.40 | Improved |
| add Phase 3 momentum fusion (ablation) | Proteus · RAF-DB | transfer ASR (%): 83.17 -> 85.40 | Improved |
| switch aggregation from FedAvg to FedSGD | FL training · RAF-DB | Proteus transfer ASR (%): 85.40 -> 77.38 | Hurt |
| apply differential privacy defense | FL model · RAF-DB | Proteus transfer ASR (%): 85.40 -> 59.64 | Hurt |
| apply differential privacy defense, clean accuracy | FL model · RAF-DB | clean accuracy (%): 71.58 -> 67.87 | Hurt |
| apply DiffPure purification | FL model · RAF-DB | clean accuracy (%): 71.58 -> 22.27 | Hurt |
| increase attack iterations from 1 to 20 (ablation) | Proteus · RAF-DB | % matching final-model mispredictions: 8.96 -> 66.67 | Improved |
| increase number of transformations beyond ~20 (ablation) | Proteus · RAF-DB, CIFAR-10, TinyImageNet | transfer ASR | No clear effect |
Closest papers by meaning
- Federated Adversarial Learning: A Framework with Convergence Analysis (ICML 2023 · similarity 0.87)
- Federated Robustness Propagation: Sharing Adversarial Robustness in Heterogeneous Federated Learning (AAAI 2023 · similarity 0.87)
- The Impact of Adversarial Attacks on Federated Learning: A Survey (TPAMI 2023 · similarity 0.87)
- FedREDefense: Defending against Model Poisoning Attacks for Federated Learning using Model Update Reconstruction Error (ICML 2024 · similarity 0.85)
- Leak and Learn: An Attacker's Cookbook to Train Using Leaked Data from Federated Learning (CVPR 2024 · similarity 0.85)
- Theoretically Unmasking Inference Attacks Against LDP-Protected Clients in Federated Vision Models (ICML 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.
