Research map / Learning theory and algorithms / Prediction and decision losses
Distribution-Free Statistical Dispersion Control for Societal Applications
Zhun Deng, Thomas P Zollo, Jake Snell, Toniann Pitassi, Richard Zemel
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 |
|---|---|---|---|
| Numerical optimization (NN-Opt.) instead of Berk-Jones bound (ablation) | bound method · CivilComments group objective n=100 | bound on max difference in group medians: 0.07549 -> 0.01612 | Improved |
| NN-Opt. instead of Berk-Jones (ablation) | bound method · CivilComments group objective n=100 | total objective bound: 0.42081 -> 0.34281 | Improved |
| NN-Opt. instead of Berk-Jones (ablation) | bound method · CivilComments group objective n=200 | total objective bound: 0.31831 -> 0.30911 | Improved |
| NN-Opt. instead of Berk-Jones (ablation) | bound method · CivilComments group objective n=200 | bound on max difference in group medians: 0.00666 -> 0.00292 | Improved |
| NN-Opt. instead of DKW (ablation) | bound method · CivilComments group objective n=100 | total objective bound: 1.27645 -> 0.34281 | Improved |
| Increase samples per group from 100 to 200 | NN-Opt. · CivilComments group objective | total objective bound: 0.34281 -> 0.30911 | Improved |
| NN-Opt. instead of Berk-Jones for CVaR (ablation) | bound method · CivilComments black female n=100 | CVaR bound: 0.91166 -> 0.85549 | Improved |
| NN-Opt. instead of Truncated Berk-Jones for CVaR (ablation) | bound method · CivilComments black female n=100 | CVaR bound: 0.86379 -> 0.85549 | Improved |
| NN-Opt. instead of Berk-Jones for VaR-Interval (ablation) | bound method · CivilComments black female n=100 | VaR-Interval bound: 0.38057 -> 0.32656 | Improved |
| NN-Opt. instead of Berk-Jones for smoothed median (ablation) | bound method · CivilComments black female n=100 | Smoothed-Median bound: 0.00038 -> 0.00021 | Improved |
| NN-Opt. adapted to CVaR instead of Berk-Jones, bound on expected loss (ablation) | bound method · CivilComments black female n=100 | expected loss bound: 0.25 -> 0.23 | Improved |
| Truncated Berk-Jones instead of Berk-Jones, expected loss bound (ablation) | Berk-Jones · CivilComments black female n=100 | expected loss bound: 0.25 -> 0.50 | Hurt |
Closest papers by meaning
- The Stability of Online Algorithms in Performative Prediction (NeurIPS 2026 · similarity 0.85)
- A Comprehensive View of Fairness through Distributional Stability (NeurIPS 2026 · similarity 0.84)
- Performative Risk Control: Calibrating Models for Reliable Deployment under Performativity (NeurIPS 2025 · similarity 0.83)
- Performative Prediction with Selective Labels (NeurIPS 2026 · similarity 0.83)
- On Statistical Rates of Conditional Diffusion Transformers: Approximation, Estimation and Minimax Optimality (ICLR 2025 · similarity 0.83)
- Group Distributionally Robust Dataset Distillation with Risk Minimization (ICLR 2025 · 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.
