Research map / Learning theory and algorithms / Fairness
On the Fairness of Causal Algorithmic Recourse
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 |
|---|---|---|---|
| Restrict classifier inputs to non-descendants of the protected attribute | Synthetic causal recourse classifiers · Synthetic IMF, CAU-LIN, and CAU-ANM | Δ_ind: Δ_ind=0 throughout | Helped |
| Restrict classifier inputs to non-descendants of the protected attribute | Synthetic causal recourse classifiers · Synthetic IMF, CAU-LIN, and CAU-ANM | accuracy: 65.3 versus 86.7; 67.3 versus 89.5; 65.6 versus 87.8 | Hurt |
| Add inferred exogenous variables for protected-attribute descendants while retaining non-descendants (ablation) | Causal recourse classifiers · Synthetic IMF, CAU-LIN, and CAU-ANM | accuracy: 65.3 -> 86.7; 67.3 -> 89.5; 65.6 -> 87.8 | Helped |
| Train a FairSVM to equalize average distance to the decision boundary | FairSVM · Synthetic IMF, linear labels | Δ_dist: 0.48 -> 0.04 | Helped |
| Train a FairSVM to equalize average distance to the decision boundary | FairSVM · Synthetic IMF, linear labels | accuracy: 86.7 -> 68.1 | Hurt |
| Use only non-protected features instead of protected and non-protected features (ablation) | LR and SVM baselines · Synthetic datasets | accuracy | No clear effect |
| Measure causal recourse cost rather than distance to the decision boundary | Recourse fairness evaluation · Synthetic CAU datasets | recourse fairness | Helped |
| Replace the true SCM with a linear SCM estimate when the true relationships are nonlinear (ablation) | Causal recourse evaluation · Synthetic CAU-ANM | Δ_cost and Δ_ind estimation | Hurt |
| Use a causal SCM to compute recourse actions rather than treating features as independently manipulable | Recourse fairness metrics · Adult | Δ_cost: Δ_dist=0.89; Δ_cost=33.32 | Helped |
| Apply a subsidy to disadvantaged-group individuals below a threshold | Credit-card toy example · Different-variance groups | difference in average recourse cost | Helped |
Closest papers by meaning
- Reconciling Predictive and Statistical Parity: A Causal Approach (AAAI 2024 · similarity 0.89)
- Causal Fairness under Unobserved Confounding: A Neural Sensitivity Framework (ICLR 2024 · similarity 0.88)
- Fairness-Accuracy Trade-Offs: A Causal Perspective (AAAI 2025 · similarity 0.88)
- Arbitrariness and Social Prediction: The Confounding Role of Variance in Fair Classification (AAAI 2024 · similarity 0.88)
- FairPFN: A Tabular Foundation Model for Causal Fairness (ICML 2025 · similarity 0.88)
- Causal Bias Detection in Generative Artifical Intelligence (NeurIPS 2026 · similarity 0.87)
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.