Research map / Federated learning and privacy / Differential privacy
GlucoSynth: Generating Differentially-Private Synthetic Glucose Traces
Josephine Lamp, Mark Derdzinski, Christopher Hannemann, Joost Van der Linden, Lu Feng, Tianhao Wang, David Evans
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 |
|---|---|---|---|
| GlucoSynth vs TimeGAN (nonprivate) | TimeGAN · glucose traces, epsilon=inf | forecasting RMSE: 0.061 -> 7e{-3} | Improved |
| GlucoSynth vs FF (nonprivate) | FF · glucose traces, epsilon=inf | validation motif fraction VM: 0.642 -> 0.987 | Improved |
| GlucoSynth vs NVP (nonprivate) | NVP · glucose traces, epsilon=inf | motif distribution MSE: 1.9 -> 1.6 | Mixed |
| GlucoSynth vs dpGAN at epsilon=1 | dpGAN · glucose traces, epsilon=1 | forecasting RMSE: 0.030 -> 0.030 | No clear effect |
| GlucoSynth vs RGAN at epsilon=1 | RGAN · glucose traces, epsilon=1 | forecasting RMSE: 0.651 -> 0.030 | Improved |
| decrease privacy budget from infinity to 0.01 | GlucoSynth · glucose traces | forecasting RMSE: 7e{-3} -> 0.038 | Hurt |
| decrease privacy budget from infinity to 0.01 | GlucoSynth · glucose traces | motif distribution MSE: 1.6 -> 99.0 | Hurt |
| decrease privacy budget from infinity to 0.01 | GlucoSynth · glucose traces | coverage: 0.534 -> 0.010 | Hurt |
| decrease epsilon from 10 to 0.1 | GlucoSynth · glucose traces | VM: 1.000 -> 1.000 | No clear effect |
| use PATE instead of DP-SGD for motif causality block | GlucoSynth · glucose traces | privacy-utility trade-off | Improved |
| GlucoSynth vs dpGAN coverage at low epsilon | dpGAN · glucose traces, epsilon=0.01 | coverage: 0.054 -> 0.010 | Hurt |
| GlucoSynth vs dpGAN motif MSE at epsilon=1 | dpGAN · glucose traces, epsilon=1 | motif distribution MSE: 23.2 -> 6.7 | Improved |
Closest papers by meaning
- PATE-GAN: Generating Synthetic Data with Differential Privacy Guarantees (ICLR 2019 · similarity 0.86)
- GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generators (NeurIPS 2020 · similarity 0.86)
- GlucoFM-Bench: Benchmarking Time-Series Foundation Models for Blood Glucose Forecasting (NeurIPS 2026 · similarity 0.83)
- Privacy without Noisy Gradients: Slicing Mechanism for Generative Model Training (NeurIPS 2024 · similarity 0.82)
- Privacy Amplification Through Synthetic Data: Insights from Linear Regression (ICML 2025 · similarity 0.82)
- A Privacy-Preserving Framework for Generative Model-driven Synthetic Datasets (AAAI 2025 · similarity 0.81)
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.
