Research map / Federated learning and privacy / Differential privacy
Fast and Memory Efficient Differentially Private-SGD via JL Projections
Zhiqi Bu, Sivakanth Gopi, Janardhan Kulkarni, Yin Tat Lee, Judy Hanwen Shen, Uthaipon Tantipongpipat
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 JL-projected gradient norms instead of exact per-sample norms | DP-Adam on IMDb · IMDb | seconds per epoch: 19317 -> 675 | Improved |
| Use JL-projected gradient norms instead of exact per-sample norms | DP-SGD · MNIST, CNN with 26,010 parameters | maximum batch size: 512 -> 60,000 | Improved |
| Use JL-projected gradient norms instead of exact per-sample norms | DP-SGD · MNIST, CNN with 17,146,938 parameters | maximum batch size: 100 -> 28,000 | Improved |
| Increase JL dimension from 1 to 30 (ablation) | DP-Adam-JL · IMDb | seconds per epoch: 41 -> 675 | Hurt |
| Increase JL dimension from 1 to 30 (ablation) | DP-Adam-JL · IMDb | privacy guarantee | Improved |
| Increase JL dimension from 3 to 20 (ablation) | DP-SGD-JL · MNIST | test accuracy | Improved |
| Use DP-Adam-JL(1) instead of DP-Adam-Vanilla (ablation) | DP-Adam · IMDb | privacy guarantee | Hurt |
Closest papers by meaning
- Private Zeroth-Order Optimization with Public Data (NeurIPS 2025 · similarity 0.88)
- Differentially private optimization for non-decomposable objective functions (ICLR 2025 · similarity 0.88)
- Differentially Private Stochastic Gradient Descent with Fixed-Size Minibatches: Tighter RDP Guarantees with or without Replacement (NeurIPS 2024 · similarity 0.88)
- Differentially Private SGD Without Clipping Bias: An Error-Feedback Approach (ICLR 2024 · similarity 0.87)
- DP-SGD Without Clipping: The Lipschitz Neural Network Way (ICLR 2024 · similarity 0.87)
- Differentially Private Prototypes for Imbalanced Transfer Learning (AAAI 2025 · 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.
