Research map / Federated learning and privacy / Differential privacy
Deep Learning with Label Differential Privacy
Badih Ghazi, Noah Golowich, Ravi Kumar, Pasin Manurangsi, Chiyuan 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 |
|---|---|---|---|
| two-stage instead of one-stage training | LP · CIFAR-10 eps=1 | test accuracy: 59.96 -> 63.67 | Improved |
| two-stage instead of one-stage training | LP · CIFAR-10 eps=2 | test accuracy: 82.38 -> 86.05 | Improved |
| two-stage instead of one-stage training | LP · CIFAR-10 eps=3 | test accuracy: 89.89 -> 92.19 | Improved |
| two-stage instead of one-stage training | LP · CIFAR-10 eps=8 | test accuracy: 94.70 -> 94.52 | Hurt |
| two-stage instead of one-stage training | LP · CIFAR-100 eps=3 | test accuracy: 20.96 -> 28.74 | Improved |
| two-stage instead of one-stage training | LP · MovieLens-1M eps=1 | test RMSE: 1.122 -> 1.034 | Improved |
| two-stage instead of one-stage training | LP · Fashion MNIST eps=1 | test accuracy: 80.78 -> 83.26 | Improved |
| two-stage instead of one-stage training | LP · KMNIST eps=1 | test accuracy: 76.56 -> 81.26 | Improved |
| add CIFAR-100 pre-training | LP-1ST · CIFAR-10 eps=1 | test accuracy: 59.96 -> 67.64 | Improved |
| add CIFAR-100 pre-training | LP-2ST · CIFAR-10 eps=8 | test accuracy: 94.52 -> 94.59 | Improved |
| go from 2 to 3 stages | LP multi-stage · CIFAR-10 and other datasets | test accuracy: 1--2% | Improved |
| go beyond 4 stages | LP multi-stage · some datasets | test accuracy | No clear effect |
Closest papers by meaning
- Differentially Private Analysis for Binary Response Models: Optimality, Estimation, and Inference (ICML 2025 · similarity 0.89)
- Enhancing Learning with Label Differential Privacy by Vector Approximation (ICLR 2025 · similarity 0.89)
- Differentially Private Sparse Reward Estimation with Preference Feedback (NeurIPS 2026 · similarity 0.85)
- Differentially Private Model Merging (NeurIPS 2026 · similarity 0.85)
- Auditing Privacy Mechanisms via Label Inference Attacks (NeurIPS 2024 · similarity 0.85)
- Differentially Private Stochastic Gradient Descent with Fixed-Size Minibatches: Tighter RDP Guarantees with or without Replacement (NeurIPS 2024 · similarity 0.85)
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.
