Research map / Learning theory and algorithms / Fairness
Eliminating Latent Discrimination: Train Then Mask
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 |
|---|---|---|---|
| Train on the sensitive feature, then mask it at prediction and tune τ | SVM · Adult Income | LDis: 0.016 -> 0.000 | Helped |
| Train on the sensitive feature, then mask it at prediction and tune τ | SVM · German Credit | LDis: 0.016 -> 0.000 | Helped |
| Train on the sensitive feature, then mask it at prediction and tune τ | SVM · COMPAS Recidivism | LDis: 0.005 -> 0.000 | Helped |
| Train on the sensitive feature, then mask it at prediction and tune τ | NN · Adult Income | LDis: 0.083 -> 0.058 | Helped |
| Train on the sensitive feature, then mask it at prediction and tune τ | NN · German Credit | LDis: 0.325 -> 0.264 | Helped |
| Train on the sensitive feature, then mask it at prediction and tune τ | NN · COMPAS Recidivism | LDis: 0.095 -> 0.076 | Helped |
| Remove the sensitive feature before training instead of training with it and masking afterward (ablation) | Train-then-mask SVM · Adult Income | Acc.: 0.823 -> 0.824 | Hurt |
| Remove the sensitive feature before training instead of training with it and masking afterward (ablation) | Train-then-mask SVM · German Credit | Acc.: 0.74 -> 0.73 | Hurt |
| Remove the sensitive feature before training instead of training with it and masking afterward (ablation) | Train-then-mask SVM · COMPAS Recidivism | Acc.: 0.766 -> 0.765 | Hurt |
Closest papers by meaning
- Why Is My Classifier Discriminatory? (NeurIPS 2018 · similarity 0.89)
- Residual Unfairness in Fair Machine Learning from Prejudiced Data (ICML 2018 · similarity 0.89)
- Reconciling Predictive and Statistical Parity: A Causal Approach (AAAI 2024 · similarity 0.88)
- Unlocking Fairness: a Trade-off Revisited (NeurIPS 2019 · similarity 0.88)
- Understanding Fairness and Prediction Error through Subspace Decomposition and Influence Analysis (NeurIPS 2025 · similarity 0.88)
- A Sequentially Fair Mechanism for Multiple Sensitive Attributes (AAAI 2024 · similarity 0.88)
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.