Research map / Federated learning and privacy / Privacy attacks and secure learning
Diverse Community Data for Benchmarking Data Privacy Algorithms
Aniruddha Sen, Christine Task, Dhruv Kapur, Gary Stanley Howarth, Karan Bhagat
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 |
|---|---|---|---|
| Add k-anonymity protection to DP synthesis (PACSynth vs MST, both eps=10) | SmartNoise DP synthesizer · CRC archive, National Excerpts | Utility (ES): =10\% -> \$\sim\$1\% | Hurt |
| Add k-anonymity protection to DP synthesis (PACSynth vs MST) | SmartNoise DP synthesizer · CRC archive, National Excerpts | Privacy leak (UEM): 13.6\% -> 0.87\% | Improved |
| Use non-DP synthetic CART instead of DP MST | synthetic data generator · CRC archive, National Excerpts | Privacy leak (UEM): 13.6\% -> 2.54\% | Hurt |
| Use non-DP CART instead of DP MST | synthetic data generator · CRC archive, National Excerpts | Utility (ES): =10\% -> \$\sim\$40\% | Improved |
| Simple DP histogram at eps=10 versus other eps=10 DP methods | DP Histogram · CRC archive, National Excerpts | Privacy leak (UEM): 100\% | Hurt |
| Same formal DP guarantee (eps=10) across techniques | DP methods · CRC archive, National Excerpts | k-marginal score: 988 -> 969 -> 551 | Mixed |
| Increase subpopulation dispersal (add features) | deidentification algorithms · CRC archive, National Excerpts | 4-marginal subgroup score | Mixed |
| CTGAN vs ADSGAN neural synthesizers | GAN synthesizers · CRC archive, National Excerpts | k-marginal score: 775 -> 121 | Mixed |
Closest papers by meaning
- The State of Data Curation at NeurIPS: An Assessment of Dataset Development Practices in the Datasets and Benchmarks Track (NeurIPS 2024 · similarity 0.83)
- Underestimated Privacy Risks for Minority Populations in Large Language Model Unlearning (ICML 2025 · similarity 0.83)
- A Simple and Practical Method for Reducing the Disparate Impact of Differential Privacy (AAAI 2024 · similarity 0.83)
- Leveraging Randomness in Model and Data Partitioning for Privacy Amplification (ICML 2025 · similarity 0.82)
- DECO-Bench: Unified Benchmark for Decoupled Task-Agnostic Synthetic Data Release (NeurIPS 2024 · similarity 0.82)
- Breaking the Quality–Privacy Tradeoff in Tabular Data Generation via In-Context Learning (NeurIPS 2026 · similarity 0.82)
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.
