Research map / Federated learning and privacy / Differential privacy
Understanding Gradient Clipping in Private SGD: A Geometric Perspective
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Gaussian perturbation before clipping per-sample gradients | DP-SGD with gradient clipping · synthetic quadratic examples (Examples 1, 2 and a 10-d Gaussian mixture) | distance to optimum ||x_t - argmin f|| | Helped |
| train progressively so gradient distribution evolves from initialization to later epochs | DP-SGD on CNNs · MNIST and CIFAR-10 gradient distributions | symmetry of projected gradient distribution | Helped |
Closest papers by meaning
- Differentially Private SGD Without Clipping Bias: An Error-Feedback Approach (ICLR 2024 · similarity 0.92)
- GeoClip: Geometry-Aware Clipping for Differentially Private SGD (NeurIPS 2025 · similarity 0.91)
- Clipped SGD Algorithms for Performative Prediction: Tight Bounds for Stochastic Bias and Remedies (ICML 2025 · similarity 0.89)
- To Clip or not to Clip: the Dynamics of SGD with Gradient Clipping in High-Dimensions (ICLR 2025 · similarity 0.89)
- Online Sensitivity Optimization in Differentially Private Learning (AAAI 2024 · similarity 0.89)
- DP-SGD Without Clipping: The Lipschitz Neural Network Way (ICLR 2024 · similarity 0.88)
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.