Group Fairness Refocused: Assessing the Social Impact of ML Systems
Uložené v:
| Názov: | Group Fairness Refocused: Assessing the Social Impact of ML Systems |
|---|---|
| Autori: | Hertweck, Corinna, Loi, Michele, Heitz, Christoph |
| Prispievatelia: | University of Zurich, Hertweck, Corinna |
| Zdroj: | 2024 11th IEEE Swiss Conference on Data Science (SDS). :189-196 |
| Informácie o vydavateľovi: | IEEE, 2024. |
| Rok vydania: | 2024 |
| Predmety: | 100 Philosophy, Fairness, 10009 Department of Informatics, 1702 Artificial Intelligence, Harm, 170: Ethik, Benefit, 000 Computer science, knowledge & systems, 006: Spezielle Computerverfahren, 1710 Information Systems, Impact, Utility, Discrimination, 1802 Information Systems and Management |
| Popis: | Fairness as a property of a prediction-based decision system is a question of its impact on the lives of affected people, which is only partially captured by standard fairness metrics. In this paper, we present a formal framework for the impact assessment of prediction-based decision systems based on the paradigm of group fairness. We generalize the equality requirements of standard fairness criteria to the concept of equality of expected impact, and we show that standard fairness criteria can be interpreted as special cases of this generalization. Furthermore, we provide a systematic and practical method for determining the necessary utility functions for modeling the impact. We conclude with a discussion of possible extensions of our approach. |
| Druh dokumentu: | Article Conference object Other literature type |
| Popis súboru: | SDS_24.pdf - application/pdf |
| DOI: | 10.1109/sds60720.2024.00034 |
| DOI: | 10.21256/zhaw-32017 |
| DOI: | 10.5167/uzh-265825 |
| Rights: | STM Policy #29 CC BY |
| Prístupové číslo: | edsair.doi.dedup.....ae03be39bbef59eea7cb47b43d4b0c9d |
| Databáza: | OpenAIRE |
Nájsť tento článok vo Web of Science | Abstrakt: | Fairness as a property of a prediction-based decision system is a question of its impact on the lives of affected people, which is only partially captured by standard fairness metrics. In this paper, we present a formal framework for the impact assessment of prediction-based decision systems based on the paradigm of group fairness. We generalize the equality requirements of standard fairness criteria to the concept of equality of expected impact, and we show that standard fairness criteria can be interpreted as special cases of this generalization. Furthermore, we provide a systematic and practical method for determining the necessary utility functions for modeling the impact. We conclude with a discussion of possible extensions of our approach. |
|---|---|
| DOI: | 10.1109/sds60720.2024.00034 |