Research map / Learning theory and algorithms / Fairness
Long-term Fairness with Selective Labels
Giovani Valdrighi, Isabel Valera, Marcos M. Raimundo
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 SELLF instead of PPO | PPO · Lending (equality of opportunity) | True disparity: 0.38 -> 0.03 | Improved |
| use SELLF instead of PPO | PPO · Lending (equality of opportunity) | Reward: 1624.64 -> 1246.24 | Hurt |
| use SELLF instead of POCAR (Oracle) | POCAR (Oracle) · Lending (equality of opportunity) | Reward: 1156.82 -> 1246.24 | Improved |
| use SELLF instead of PPO | PPO · Criminal recidivism (accuracy parity) | True disparity: 0.09 -> 0.05 | Improved |
| use SELLF instead of PPO | PPO · Criminal recidivism (accuracy parity) | Reward: 998.61 -> 990.72 | No clear effect |
| use SELLF instead of POCAR | POCAR · Criminal recidivism (accuracy parity) | Reward: 838.94 -> 990.72 | Improved |
| use SELLF instead of PPO | PPO · School admission (qualification parity) | True disparity: 0.15 -> 0.12 | Improved |
| use SELLF instead of PPO | PPO · School admission (qualification parity) | Reward: 1219.68 -> 1131.58 | Hurt |
| use semi-stochastic policy SELLF (reject if pi<0.25) instead of deterministic SELLF | SELLF · Lending (equality of opportunity) | True disparity: 0.03 -> 0.05 | Hurt |
| use semi-stochastic policy SELLF instead of SELLF | SELLF · Criminal recidivism (accuracy parity) | True disparity: 0.05 -> 0.02 | Improved |
| use semi-stochastic policy SELLF instead of SELLF | SELLF · Criminal recidivism (accuracy parity) | Reward: 990.72 -> 956.07 | Hurt |
| add Renyi regularization with beta2 >= 0.1 (ablation) | SELLF · Lending (accuracy parity) | Disparity gap: beta2=0.1 and 0.2 minimize gap, beta2 < 0.1 gap above 0.01 | Improved |
Closest papers by meaning
- A Causal Lens for Learning Long-term Fair Policies (ICLR 2025 · similarity 0.89)
- Adapting Static Fairness to Sequential Decision-Making: Bias Mitigation Strategies towards Equal Long-term Benefit Rate (ICML 2024 · similarity 0.87)
- Remembering to Be Fair: Non-Markovian Fairness in Sequential Decision Making (ICML 2024 · similarity 0.86)
- Learning Fair Policies for Multi-Stage Selection Problems from Observational Data (AAAI 2024 · similarity 0.86)
- Performative Prediction with Selective Labels (NeurIPS 2026 · similarity 0.85)
- Fair Off-Policy Learning from Observational Data (ICML 2024 · similarity 0.85)
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.
