Research map / Adversarial robustness and security / Attacks on text and graphs
MExMI: Pool-based Active Model Extraction Crossover Membership Inference
Yaxin Xiao, Qingqing Ye, Haibo Hu, Huadi Zheng, Chengfang Fang, Jie Shi
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 MI post-filter to ActiveThief (ablation) | ActiveThief PAME · CIFAR10 | fidelity: 1.64% increase | Improved |
| add MI post-filter (ablation) | ActiveThief PAME · AG's News | fidelity: 1.57% | Improved |
| add MI pre-filter (greedy k-center) (ablation) | ActiveThief PAME · CIFAR10 Greedy K-center | fidelity: 86.26 -> 85.84 | Hurt |
| add MI pre-filter (entropy uncertainty) (ablation) | ActiveThief PAME · CIFAR10 Entropy Uncertainty | fidelity: 84.65 -> 85.38 | Improved |
| combine pre-filter and post-filter (ablation) | ActiveThief PAME · CIFAR10 Entropy Uncertainty | fidelity: 84.65 -> 89.10 | Improved |
| add semi-supervised boosting (ablation) | MExMI · CIFAR10 Entropy Uncertainty | fidelity: 89.10 -> 92.41 | Improved |
| add semi-supervised boosting (ablation) | MExMI · AG's News Entropy Uncertainty | fidelity: 85.18 -> 85.82 | Improved |
| remove victim training data from pool (0%) | MExMI vs baseline · CIFAR10 | fidelity: 76.30 -> 79.11 | Improved |
| pre-filter only with no training data in pool (ablation) | Pre-Filter only · CIFAR10 0% pool | fidelity: 76.30 -> 74.23 | Hurt |
| raise training data share of pool to 25% | MExMI · CIFAR10 | fidelity: 79.11 -> 92.91 | Improved |
| limit output access to top-1 score | MExMI · CIFAR10 | fidelity: 94.07 -> 91.00 | Hurt |
| homogeneous vs non-homogeneous pool | ActiveThief · CIFAR10 | fidelity: 90.32 -> 92.30 | Improved |
Closest papers by meaning
- HoneypotNet: Backdoor Attacks Against Model Extraction (AAAI 2025 · similarity 0.84)
- Stealix: Model Stealing via Prompt Evolution (ICML 2025 · similarity 0.83)
- From Counterfactuals to Trees: Competitive Analysis of Model Extraction Attacks (NeurIPS 2025 · similarity 0.83)
- Model Inversion Robustness: Can Transfer Learning Help? (CVPR 2024 · similarity 0.83)
- Chameleon: Increasing Label-Only Membership Leakage with Adaptive Poisoning (ICLR 2024 · similarity 0.82)
- SAME: Sample Reconstruction against Model Extraction Attacks (AAAI 2024 · similarity 0.82)
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.
