Post-processing fairness with minimal changes
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866929478498779136 |
|---|---|
| author | Di Gennaro, Federico Laugel, Thibault Grari, Vincent Renard, Xavier Detyniecki, Marcin |
| author_facet | Di Gennaro, Federico Laugel, Thibault Grari, Vincent Renard, Xavier Detyniecki, Marcin |
| contents | In this paper, we introduce a novel post-processing algorithm that is both model-agnostic and does not require the sensitive attribute at test time. In addition, our algorithm is explicitly designed to enforce minimal changes between biased and debiased predictions; a property that, while highly desirable, is rarely prioritized as an explicit objective in fairness literature. Our approach leverages a multiplicative factor applied to the logit value of probability scores produced by a black-box classifier. We demonstrate the efficacy of our method through empirical evaluations, comparing its performance against other four debiasing algorithms on two widely used datasets in fairness research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_15096 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Post-processing fairness with minimal changes Di Gennaro, Federico Laugel, Thibault Grari, Vincent Renard, Xavier Detyniecki, Marcin Machine Learning Artificial Intelligence In this paper, we introduce a novel post-processing algorithm that is both model-agnostic and does not require the sensitive attribute at test time. In addition, our algorithm is explicitly designed to enforce minimal changes between biased and debiased predictions; a property that, while highly desirable, is rarely prioritized as an explicit objective in fairness literature. Our approach leverages a multiplicative factor applied to the logit value of probability scores produced by a black-box classifier. We demonstrate the efficacy of our method through empirical evaluations, comparing its performance against other four debiasing algorithms on two widely used datasets in fairness research. |
| title | Post-processing fairness with minimal changes |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2408.15096 |