Explainable post-training bias mitigation with distribution-based fairness metrics
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909877139406848 |
|---|---|
| author | Franks, Ryan Miroshnikov, Alexey Kotsiopoulos, Konstandinos |
| author_facet | Franks, Ryan Miroshnikov, Alexey Kotsiopoulos, Konstandinos |
| contents | We develop a novel bias mitigation framework with distribution-based fairness constraints suitable for producing demographically blind and explainable machine-learning models across a wide range of fairness levels. This is accomplished through post-processing, allowing fairer models to be generated efficiently without retraining the underlying model. Our framework, which is based on stochastic gradient descent, can be applied to a wide range of model types, with a particular emphasis on the post-processing of gradient-boosted decision trees. Additionally, we design a broad family of global fairness metrics, along with differentiable and consistent estimators compatible with our framework, building on previous work. We empirically test our methodology on a variety of datasets and compare it with alternative post-processing approaches, including Bayesian search, optimal transport projection, and direct neural network training. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_01223 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Explainable post-training bias mitigation with distribution-based fairness metrics Franks, Ryan Miroshnikov, Alexey Kotsiopoulos, Konstandinos Machine Learning Probability 49Q22, 65K10, 91A12, 68T01 We develop a novel bias mitigation framework with distribution-based fairness constraints suitable for producing demographically blind and explainable machine-learning models across a wide range of fairness levels. This is accomplished through post-processing, allowing fairer models to be generated efficiently without retraining the underlying model. Our framework, which is based on stochastic gradient descent, can be applied to a wide range of model types, with a particular emphasis on the post-processing of gradient-boosted decision trees. Additionally, we design a broad family of global fairness metrics, along with differentiable and consistent estimators compatible with our framework, building on previous work. We empirically test our methodology on a variety of datasets and compare it with alternative post-processing approaches, including Bayesian search, optimal transport projection, and direct neural network training. |
| title | Explainable post-training bias mitigation with distribution-based fairness metrics |
| topic | Machine Learning Probability 49Q22, 65K10, 91A12, 68T01 |
| url | https://arxiv.org/abs/2504.01223 |