Research map / Learning theory and algorithms / Matrix and tensor methods
Selective Inference with Distributed Data
Sifan Liu, Snigdha Panigrahi
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 |
|---|---|---|---|
| reuse local-machine data for selective inference instead of sample splitting | Dist-SI · GOSSIS ICU diabetes | significant variables: 13 -> 21 | Improved |
| reuse local-machine data for selective inference instead of sample splitting | Dist-SI · GOSSIS ICU diabetes | median CI length: 67% smaller median interval length | Improved |
| reuse local-machine data instead of sample splitting | Dist-SI · GOSSIS ICU diabetes | coefficient of variation of interval length: 3.9 -> 1.8 | Improved |
| Dist-SI instead of sample splitting | Dist-SI · simulated distributed regression (Scenarios 1-3) | CI length and power: tighter intervals and higher power | Improved |
| increase number of machines K from 2 to 8 (ablation) | Dist-SI and Splitting · Scenario 1 simulation | CI length: K in {2,4,6,8} | Hurt |
| increase signal strength c (ablation) | Dist-SI and Splitting · Scenario 2 simulation | power: c from 0.3 to 0.9 | Improved |
| decrease central-machine samples n0 (ablation) | Dist-SI and Splitting · Scenario 3 simulation | CI length: n0 from 2000 to 250 | Hurt |
| Dist-SI instead of Multi-carving / Multi-splitting | Dist-SI · P-value lottery simulation | diagnostic odds ratio: larger DOR for all sample proportions | Improved |
| Dist-SI instead of Multi-carving | Dist-SI · P-value lottery simulation | run time: about 100 times faster | Improved |
Closest papers by meaning
- Selective inference for group-sparse linear models (NeurIPS 2016 · similarity 0.89)
- Communication-efficient Sparse Regression (JMLR 2017 · similarity 0.88)
- Debiased Distributed Learning for Sparse Partial Linear Models in High Dimensions (JMLR 2022 · similarity 0.88)
- Feature-distributed sparse regression: a screen-and-clean approach (NeurIPS 2016 · similarity 0.87)
- Acceleration in Distributed Sparse Regression (NeurIPS 2022 · similarity 0.86)
- Better Locally Private Sparse Estimation Given Multiple Samples Per User (ICML 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.
