Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Hertweck, Corinna | - |
dc.contributor.author | Räz, Tim | - |
dc.date.accessioned | 2023-12-15T08:46:40Z | - |
dc.date.available | 2023-12-15T08:46:40Z | - |
dc.date.issued | 2022-02-25 | - |
dc.identifier.isbn | 978-1-57735-876-3 | de_CH |
dc.identifier.issn | 2374-3468 | de_CH |
dc.identifier.issn | 2159-5399 | de_CH |
dc.identifier.uri | https://digitalcollection.zhaw.ch/handle/11475/29379 | - |
dc.description | AAM of the article with technical appendix can be found here: https://arxiv.org/abs/2109.04399 | de_CH |
dc.description.abstract | Impossibility results show that important fairness measures (independence, separation, sufficiency) cannot be satisfied at the same time under reasonable assumptions. This paper explores whether we can satisfy and/or improve these fairness measures simultaneously to a certain degree. We introduce information-theoretic formulations of the fairness measures and define degrees of fairness based on these formulations. The information-theoretic formulations suggest unexplored theoretical relations between the three fairness measures. In the experimental part, we use the information-theoretic expressions as regularizers to obtain fairness-regularized predictors for three standard datasets. Our experiments show that a) fairness regularization directly increases fairness measures, in line with existing work, and b) some fairness regularizations indirectly increase other fairness measures, as suggested by our theoretical findings. This establishes that it is possible to increase the degree to which some fairness measures are satisfied at the same time -- some fairness measures are gradually compatible. | de_CH |
dc.language.iso | en | de_CH |
dc.publisher | Association for the Advancement of Artificial Intelligence | de_CH |
dc.rights | Licence according to publishing contract | de_CH |
dc.subject | Fairness | de_CH |
dc.subject | Discrimination | de_CH |
dc.subject | Statistical parity | de_CH |
dc.subject | Accuracy | de_CH |
dc.subject | Mutual information | de_CH |
dc.subject | Entropy | de_CH |
dc.subject | Separation | de_CH |
dc.subject | Sufficiency | de_CH |
dc.subject | Independence | de_CH |
dc.subject.ddc | 006: Spezielle Computerverfahren | de_CH |
dc.subject.ddc | 170: Ethik | de_CH |
dc.title | Gradual (in)compatibility of fairness criteria | de_CH |
dc.type | Konferenz: Paper | de_CH |
dcterms.type | Text | de_CH |
zhaw.departement | School of Engineering | de_CH |
zhaw.organisationalunit | Institut für Datenanalyse und Prozessdesign (IDP) | de_CH |
zhaw.publisher.place | Palo Alto, CA | de_CH |
dc.identifier.doi | 10.1609/aaai.v36i11.21450 | de_CH |
zhaw.conference.details | 36th AAAI Conference on Artificial Intelligence, online, 22 February–1 March 2022 | de_CH |
zhaw.funding.eu | No | de_CH |
zhaw.issue | 11 | de_CH |
zhaw.originated.zhaw | Yes | de_CH |
zhaw.pages.end | 11934 | de_CH |
zhaw.pages.start | 11926 | de_CH |
zhaw.publication.status | publishedVersion | de_CH |
zhaw.volume | 36 | de_CH |
zhaw.publication.review | Peer review (Publikation) | de_CH |
zhaw.title.proceedings | Proceedings of the 36th AAAI Conference on Artificial Intelligence : Vol. 36 No. 11: IAAI-22, EAAI-22, AAAI-22 Special Programs and Special Track, Student Papers and Demonstrations | de_CH |
zhaw.funding.snf | 187473 | de_CH |
zhaw.funding.zhaw | Socially acceptable AI and fairness trade-offs in predictive analytics | de_CH |
zhaw.author.additional | No | de_CH |
zhaw.display.portrait | Yes | de_CH |
Appears in collections: | Publikationen School of Engineering |
Files in This Item:
There are no files associated with this item.
Show simple item record
Hertweck, C., & Räz, T. (2022). Gradual (in)compatibility of fairness criteria [Conference paper]. Proceedings of the 36th AAAI Conference on Artificial Intelligence : Vol. 36 No. 11: IAAI-22, EAAI-22, AAAI-22 Special Programs and Special Track, Student Papers and Demonstrations, 36(11), 11926–11934. https://doi.org/10.1609/aaai.v36i11.21450
Hertweck, C. and Räz, T. (2022) ‘Gradual (in)compatibility of fairness criteria’, in Proceedings of the 36th AAAI Conference on Artificial Intelligence : Vol. 36 No. 11: IAAI-22, EAAI-22, AAAI-22 Special Programs and Special Track, Student Papers and Demonstrations. Palo Alto, CA: Association for the Advancement of Artificial Intelligence, pp. 11926–11934. Available at: https://doi.org/10.1609/aaai.v36i11.21450.
C. Hertweck and T. Räz, “Gradual (in)compatibility of fairness criteria,” in Proceedings of the 36th AAAI Conference on Artificial Intelligence : Vol. 36 No. 11: IAAI-22, EAAI-22, AAAI-22 Special Programs and Special Track, Student Papers and Demonstrations, Feb. 2022, vol. 36, no. 11, pp. 11926–11934. doi: 10.1609/aaai.v36i11.21450.
HERTWECK, Corinna und Tim RÄZ, 2022. Gradual (in)compatibility of fairness criteria. In: Proceedings of the 36th AAAI Conference on Artificial Intelligence : Vol. 36 No. 11: IAAI-22, EAAI-22, AAAI-22 Special Programs and Special Track, Student Papers and Demonstrations. Conference paper. Palo Alto, CA: Association for the Advancement of Artificial Intelligence. 25 Februar 2022. S. 11926–11934. ISBN 978-1-57735-876-3
Hertweck, Corinna, and Tim Räz. 2022. “Gradual (in)Compatibility of Fairness Criteria.” Conference paper. In Proceedings of the 36th AAAI Conference on Artificial Intelligence : Vol. 36 No. 11: IAAI-22, EAAI-22, AAAI-22 Special Programs and Special Track, Student Papers and Demonstrations, 36:11926–34. Palo Alto, CA: Association for the Advancement of Artificial Intelligence. https://doi.org/10.1609/aaai.v36i11.21450.
Hertweck, Corinna, and Tim Räz. “Gradual (in)Compatibility of Fairness Criteria.” Proceedings of the 36th AAAI Conference on Artificial Intelligence : Vol. 36 No. 11: IAAI-22, EAAI-22, AAAI-22 Special Programs and Special Track, Student Papers and Demonstrations, vol. 36, no. 11, Association for the Advancement of Artificial Intelligence, 2022, pp. 11926–34, https://doi.org/10.1609/aaai.v36i11.21450.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.