Research map / Learning theory and algorithms / Fairness
Adaptive Data Debiasing through Bounded Exploration
Yifan Yang, Yang Liu, Parinaz Naghizadeh
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 |
|---|---|---|---|
| Use bounded exploration instead of exploitation-only data collection | Active debiasing · Classification with censored feedback | distribution-estimation bias | Improved |
| Use exploitation-only data collection with no exploration | Distribution estimation · Classification with censored feedback | distribution-estimation bias | Hurt |
| Explore below the decision threshold without a lower bound | Pure exploration · Classification with censored feedback | distribution-estimation bias | Improved |
| Replace unbounded pure exploration with bounded exploration | Active debiasing · Gaussian synthetic data | speed of debiasing | Hurt |
| Explore more deeply below the classifier | Active debiasing and pure exploration · Gaussian synthetic data | regret | Improved |
| Explore more deeply below the classifier | Active debiasing and pure exploration · Gaussian synthetic data | weighted regret | Hurt |
| Use a higher exploration percentile, limiting the exploration depth (ablation) | Active debiasing · Gaussian synthetic data | weighted regret | Improved |
| Apply a fairness constraint that over-selects a group | Active debiasing · Groups subject to fairness constraints | speed of debiasing | Hurt |
| Apply a fairness constraint that under-selects a group | Active debiasing · Groups subject to fairness constraints | speed of debiasing | Improved |
| Use dimension reduction to map Adult features to a one-dimensional score (ablation) | Logistic regression · Adult, overall | accuracy: 82.96% -> 78.07% | Hurt |
| Use dimension reduction to map Adult features to a one-dimensional score (ablation) | Logistic regression · Adult, advantaged group | accuracy: 82.13% -> 76.95% | Hurt |
| Use dimension reduction to map Adult features to a one-dimensional score (ablation) | Logistic regression · Adult, disadvantaged group | accuracy: 88.26% -> 84.69% | Hurt |
Closest papers by meaning
- Fair Classification with Partial Feedback: An Exploration-Based Data Collection Approach (ICML 2024 · similarity 0.86)
- Fair Off-Policy Learning from Observational Data (ICML 2024 · similarity 0.84)
- Learning to Persuade a Biased Receiver (NeurIPS 2026 · similarity 0.84)
- Improving Subgroup Robustness via Data Selection (NeurIPS 2024 · similarity 0.83)
- Post-hoc bias scoring is optimal for fair classification (ICLR 2024 · similarity 0.83)
- Fairness without Harm: An Influence-Guided Active Sampling Approach (NeurIPS 2024 · similarity 0.83)
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.
