Research map / Learning theory and algorithms / Fairness
Faking Fairness via Stealthily Biased Sampling
Kazuto Fukuchi, Satoshi Hara, Takanori Maehara
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 stealthily biased sampling instead of random sampling to construct the disclosed benchmark | Decision-maker's benchmark dataset · Synthetic loan check | demographic parity: 0.2 -> less than 0.1 | Improved |
| Use case-control sampling instead of stealthily biased sampling | Decision-maker's benchmark dataset · Synthetic loan check | demographic parity | No clear effect |
| Preserve similarity to the original feature distribution while enforcing fairness | Stealthily biased sampling · Synthetic loan check | KS-test rejection rate: around 0.05 | Improved |
| Use stealthily biased sampling rather than case-control sampling | Decision-maker's benchmark dataset · Synthetic loan check | KS-test rejection rate | Improved |
| Use stealthily biased sampling rather than case-control sampling | Decision-maker's benchmark dataset · COMPAS | Wasserstein distance | Improved |
| Use stealthily biased sampling to select the disclosed subset | COMPAS risk-score decisions · COMPAS | demographic parity | Improved |
| Use stealthily biased sampling to select the disclosed subset | Logistic-regression decisions · Adult | demographic parity | Improved |
| Use stealthily biased sampling with stochastic rather than deterministic decisions | Synthetic loan check decision-maker · Synthetic loan check | range of positive-decision ratios that evade detection: α=0.6 -> α between 0.25 and 0.95 | Improved |
Closest papers by meaning
- Cost Efficient Fairness Audit Under Partial Feedback (NeurIPS 2026 · similarity 0.84)
- Procedural Fairness Through Decoupling Objectionable Data Generating Components (ICLR 2024 · similarity 0.84)
- Fairness Issues and Mitigations in (Differentially Private) Socio-Demographic Data Processes (AAAI 2025 · similarity 0.84)
- Fairness without Harm: An Influence-Guided Active Sampling Approach (NeurIPS 2024 · similarity 0.84)
- Statistical Inference for Fairness Auditing (NeurIPS 2024 · similarity 0.83)
- Robust ML Auditing using Prior Knowledge (ICML 2025 · 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.
