Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-Leak: Membership Inference Attacks against Semi-Supervised Learning
Xinlei He, Hongbin Liu, Neil Zhenqiang Gong, Yang Zhang
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 A_DA instead of the best baseline attack A_NN | membership inference on FixMatch · CIFAR10 (500 labels) | attack AUC: 0.722 -> 0.780 | Improved |
| increase augmented views from 10 to 100 (ablation) | A_DA · CIFAR10 (FixMatch, 500 labels) | attack AUC: 0.780 -> 0.806 | Improved |
| use JS Distance instead of cosine, correlation or Euclidean distance (ablation) | A_DA · CIFAR10 (FixMatch, 500 labels) | attack AUC: 0.679, 0.682, 0.749 -> 0.780 | Improved |
| enlarge target model from WRN28-1 to WRN28-8 (ablation) | A_DA against FixMatch · CIFAR100 (500 labels) | attack AUC: 0.726 -> 0.927 | Improved |
| raise unlabeled batch ratio from 1 to 16 (ablation) | A_DA against FixMatch · CIFAR100 (500 labels) | attack AUC: 0.578 -> 0.909 | Improved |
| raise unlabeled batch ratio from 1 to 16 (ablation) | FixMatch target model · CIFAR100 (500 labels) | test accuracy: 0.210 -> 0.247 | Mixed |
| use the strongest augmentation level 4 instead of level 0 (ablation) | A_DA · CIFAR100 (FixMatch, 500 labels) | attack AUC: 0.876 -> 0.882 | Improved |
| use ResNet50 shadow model instead of same-architecture WRN28-2 (ablation) | A_DA · CIFAR100 (FixMatch, 500 labels) | attack AUC: 0.880 -> 0.859 | Hurt |
| add early stopping | FixMatch target model · CIFAR100 (4,000 labels) | attack AUC: 0.918 -> 0.695 | Improved |
| add early stopping | FixMatch target model · CIFAR100 (4,000 labels) | test accuracy: 0.530 -> 0.490 | Hurt |
| add DP-SGD | FixMatch target model · CIFAR100 (4,000 labels) | test accuracy: 0.530 -> 0.034 | Hurt |
| add top-k (k=1) posterior release | FixMatch target model · CIFAR100 (4,000 labels) | attack AUC: 0.918 -> 0.906 | No clear effect |
Closest papers by meaning
- Sequential Membership Inference Attacks (NeurIPS 2026 · similarity 0.89)
- Chameleon: Increasing Label-Only Membership Leakage with Adaptive Poisoning (ICLR 2024 · similarity 0.89)
- Low-Cost High-Power Membership Inference Attacks (ICML 2024 · similarity 0.88)
- Exploring the limits of strong membership inference attacks on large language models (NeurIPS 2025 · similarity 0.87)
- Confidence Is All You Need for MI Attacks (Student Abstract) (AAAI 2024 · similarity 0.87)
- Membership Inference Attacks against Fine-tuned Large Language Models via Self-prompt Calibration (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.
