Research map / Adversarial robustness and security / Attacks on text and graphs
Improving Robustness to Model Inversion Attacks via Mutual Information Regularization
Tianhao Wang, Yuheng Zhang, Ruoxi Jia
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 |
|---|---|---|---|
| Regularize mutual information between model input and prediction instead of using DP | MID versus DP · Blackbox MI attacks on IWPC, FiveThirtyEight, FaceScrub, and CIFAR10 | privacy-utility tradeoff | Improved |
| Regularize mutual information between model input and prediction instead of using DP | MID versus DP · Whitebox MI attacks on CelebA and FiveThirtyEight | privacy-utility tradeoff | Improved |
| Use MID rather than DP at comparable target-model performance | FaceScrub neural network · CelebA GMI image reconstruction | test accuracy: 80.8; 81.4 | No clear effect |
| Use MID rather than DP | Neural networks under Knowledge Alignment and Update-Leaks · FaceScrub and CIFAR10 | ECE | Improved |
| Use MID rather than Priority | Decision tree · FiveThirtyEight | privacy-utility tradeoff | Mixed |
| Use MID rather than output-confidence rounding or uniform-noise injection | Blackbox neural-network defenses · FaceScrub and CIFAR10 MI attacks | attack robustness | Improved |
Closest papers by meaning
- A Sample-Level Evaluation and Generative Framework for Model Inversion Attacks (AAAI 2025 · similarity 0.91)
- Model Inversion Robustness: Can Transfer Learning Help? (CVPR 2024 · similarity 0.91)
- Trap-MID: Trapdoor-based Defense against Model Inversion Attacks (NeurIPS 2024 · similarity 0.89)
- Pseudo-Private Data Guided Model Inversion Attacks (NeurIPS 2024 · similarity 0.86)
- High-Fidelity Gradient Inversion in Distributed Learning (AAAI 2024 · similarity 0.86)
- Efficient Model Stealing Defense with Noise Transition Matrix (CVPR 2024 · similarity 0.86)
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.
