Equalised Odds is not Equal Individual Odds: Post-processing for Group and Individual Fairness
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2023
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866916212965900288 |
|---|---|
| author | Small, Edward A. Sokol, Kacper Manning, Daniel Salim, Flora D. Chan, Jeffrey |
| author_facet | Small, Edward A. Sokol, Kacper Manning, Daniel Salim, Flora D. Chan, Jeffrey |
| contents | Group fairness is achieved by equalising prediction distributions between protected sub-populations; individual fairness requires treating similar individuals alike. These two objectives, however, are incompatible when a scoring model is calibrated through discontinuous probability functions, where individuals can be randomly assigned an outcome determined by a fixed probability. This procedure may provide two similar individuals from the same protected group with classification odds that are disparately different -- a clear violation of individual fairness. Assigning unique odds to each protected sub-population may also prevent members of one sub-population from ever receiving equal chances of a positive outcome to another, which we argue is another type of unfairness called individual odds. We reconcile all this by constructing continuous probability functions between group thresholds that are constrained by their Lipschitz constant. Our solution preserves the model's predictive power, individual fairness and robustness while ensuring group fairness. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2304_09779 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Equalised Odds is not Equal Individual Odds: Post-processing for Group and Individual Fairness Small, Edward A. Sokol, Kacper Manning, Daniel Salim, Flora D. Chan, Jeffrey Machine Learning Computers and Society Optimization and Control Probability Group fairness is achieved by equalising prediction distributions between protected sub-populations; individual fairness requires treating similar individuals alike. These two objectives, however, are incompatible when a scoring model is calibrated through discontinuous probability functions, where individuals can be randomly assigned an outcome determined by a fixed probability. This procedure may provide two similar individuals from the same protected group with classification odds that are disparately different -- a clear violation of individual fairness. Assigning unique odds to each protected sub-population may also prevent members of one sub-population from ever receiving equal chances of a positive outcome to another, which we argue is another type of unfairness called individual odds. We reconcile all this by constructing continuous probability functions between group thresholds that are constrained by their Lipschitz constant. Our solution preserves the model's predictive power, individual fairness and robustness while ensuring group fairness. |
| title | Equalised Odds is not Equal Individual Odds: Post-processing for Group and Individual Fairness |
| topic | Machine Learning Computers and Society Optimization and Control Probability |
| url | https://arxiv.org/abs/2304.09779 |