Research map / Learning theory and algorithms / Prediction and decision losses
Fairwashing: the risk of rationalization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Increase the fairness weight β in LaundryML-global | LaundryML-global · Adult Income and ProPublica Recidivism suing groups | unfairness | Helped |
| Increase the fairness weight β in LaundryML-global | LaundryML-global · Adult Income and ProPublica Recidivism suing groups | fidelity | Hurt |
| Decrease the rule-list length penalty λ | LaundryML-global · Adult Income and ProPublica Recidivism suing groups | fidelity | Helped |
| Increase the fairness weight β | LaundryML-local · Adult Income minority-group rejected users | agreement coverage: 92.90% -> 100% | Helped |
| Increase the fairness weight β | LaundryML-local · ProPublica Recidivism minority-group rejected users | agreement coverage: 100% -> 100% | No clear effect |
| Increase the fairness weight β | LaundryML-local · Adult Income minority-group rejected users | unfairness: unfairness = 0.0 | Helped |
| Use a smaller fairness weight when black-box unfairness is low | LaundryML-global · Adult Income, alternative fairness metrics | number of fairwashing candidate models: unfairness < 0.1; β < 0.5 | Helped |
| Use a larger fairness weight when black-box unfairness is high | LaundryML-global · Adult Income, alternative fairness metrics | number of fairwashing candidate models: unfairness ≥ 0.1; β ≥ 0.5 | Helped |
| Apply LaundryML-global rationalization to alternative fairness metrics | LaundryML-global · Adult Income suing group | availability of fairwashing candidates | Helped |
| Apply LaundryML-global rationalization to alternative black-box model types | LaundryML-global · Adult Income suing group | availability of fairwashing candidates | Helped |
Closest papers by meaning
- Counterfactual Fairness by Combining Factual and Counterfactual Predictions (NeurIPS 2024 · similarity 0.84)
- The Fragility of Fairness: Causal Sensitivity Analysis for Fair Machine Learning (NeurIPS 2024 · similarity 0.84)
- Understanding Fairness and Prediction Error through Subspace Decomposition and Influence Analysis (NeurIPS 2025 · similarity 0.84)
- Fair Off-Policy Learning from Observational Data (ICML 2024 · similarity 0.84)
- Reconciling Predictive and Statistical Parity: A Causal Approach (AAAI 2024 · similarity 0.84)
- Mind the Gap: A Causal Perspective on Bias Amplification in Prediction & Decision-Making (NeurIPS 2024 · similarity 0.84)
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.