Please use this identifier to cite or link to this item:
https://doi.org/10.21256/zhaw-23393
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Wehrli, Samuel | - |
dc.contributor.author | Hertweck, Corinna | - |
dc.contributor.author | Amirian, Mohammadreza | - |
dc.contributor.author | Glüge, Stefan | - |
dc.contributor.author | Stadelmann, Thilo | - |
dc.date.accessioned | 2021-11-03T12:39:05Z | - |
dc.date.available | 2021-11-03T12:39:05Z | - |
dc.date.issued | 2021-10-27 | - |
dc.identifier.issn | 2730-5953 | de_CH |
dc.identifier.issn | 2730-5961 | de_CH |
dc.identifier.uri | https://digitalcollection.zhaw.ch/handle/11475/23393 | - |
dc.description.abstract | Face Recognition (FR) is increasingly influencing our lives: we use it to unlock our phones; police uses it to identify suspects. Two main concerns are associated with this increase in facial recognition: (1) the fact that these systems are typically less accurate for marginalized groups, which can be described as “bias”, and (2) the increased surveillance through these systems. Our paper is concerned with the first issue. Specifically, we explore an intuitive technique for reducing this bias, namely “blinding” models to sensitive features, such as gender or race, and show why this cannot be equated with reducing bias. Even when not designed for this task, facial recognition models can deduce sensitive features, such as gender or race, from pictures of faces—simply because they are trained to determine the “similarity” of pictures. This means that people with similar skin tones, similar hair length, etc. will be seen as similar by facial recognition models. When confronted with biased decision-making by humans, one approach taken in job application screening is to “blind” the human decision-makers to sensitive attributes such as gender and race by not showing pictures of the applicants. Based on a similar idea, one might think that if facial recognition models were less aware of these sensitive features, the difference in accuracy between groups would decrease. We evaluate this assumption—which has already penetrated into the scientific literature as a valid de-biasing method—by measuring how “aware” models are of sensitive features and correlating this with differences in accuracy. In particular, we blind pre-trained models to make them less aware of sensitive attributes. We find that awareness and accuracy do not positively correlate, i.e., that bias ≠ awareness. In fact, blinding barely affects accuracy in our experiments. The seemingly simple solution of decreasing bias in facial recognition rates by reducing awareness of sensitive features does thus not work in practice: trying to ignore sensitive attributes is not a viable concept for less biased FR. | de_CH |
dc.language.iso | en | de_CH |
dc.publisher | Springer | de_CH |
dc.relation.ispartof | AI and Ethics | de_CH |
dc.rights | http://creativecommons.org/licenses/by/4.0/ | de_CH |
dc.subject | Fairness | de_CH |
dc.subject | Convolutional neural network | de_CH |
dc.subject | Discrimination | de_CH |
dc.subject | Ethnic bias | de_CH |
dc.subject | Gender bias | de_CH |
dc.subject.ddc | 006: Spezielle Computerverfahren | de_CH |
dc.subject.ddc | 170: Ethik | de_CH |
dc.title | Bias, awareness, and ignorance in deep-learning-based face recognition | de_CH |
dc.type | Beitrag in wissenschaftlicher Zeitschrift | de_CH |
dcterms.type | Text | de_CH |
zhaw.departement | Life Sciences und Facility Management | de_CH |
zhaw.departement | School of Engineering | de_CH |
zhaw.departement | Soziale Arbeit | de_CH |
zhaw.organisationalunit | Centre for Artificial Intelligence (CAI) | de_CH |
zhaw.organisationalunit | Institut für Computational Life Sciences (ICLS) | de_CH |
zhaw.organisationalunit | Institut für Datenanalyse und Prozessdesign (IDP) | de_CH |
dc.identifier.doi | 10.1007/s43681-021-00108-6 | de_CH |
dc.identifier.doi | 10.21256/zhaw-23393 | - |
zhaw.funding.eu | No | de_CH |
zhaw.issue | 3 | de_CH |
zhaw.originated.zhaw | Yes | de_CH |
zhaw.pages.end | 522 | de_CH |
zhaw.pages.start | 509 | de_CH |
zhaw.publication.status | publishedVersion | de_CH |
zhaw.volume | 2 | de_CH |
zhaw.publication.review | Peer review (Publikation) | de_CH |
zhaw.webfeed | Biosensor Analysis & Digital Health | de_CH |
zhaw.webfeed | Machine Perception and Cognition | de_CH |
zhaw.webfeed | Datalab | de_CH |
zhaw.webfeed | Predictive Analytics | de_CH |
zhaw.webfeed | ZHAW digital | de_CH |
zhaw.funding.zhaw | Libra: A One-Tool Solution for MLD4 Compliance | de_CH |
zhaw.author.additional | No | de_CH |
zhaw.display.portrait | Yes | de_CH |
Appears in collections: | Publikationen Soziale Arbeit |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
2021_Wehrli-etal_Deep-learning-face-recognition.pdf | 1.68 MB | Adobe PDF | View/Open |
Show simple item record
Wehrli, S., Hertweck, C., Amirian, M., Glüge, S., & Stadelmann, T. (2021). Bias, awareness, and ignorance in deep-learning-based face recognition. AI and Ethics, 2(3), 509–522. https://doi.org/10.1007/s43681-021-00108-6
Wehrli, S. et al. (2021) ‘Bias, awareness, and ignorance in deep-learning-based face recognition’, AI and Ethics, 2(3), pp. 509–522. Available at: https://doi.org/10.1007/s43681-021-00108-6.
S. Wehrli, C. Hertweck, M. Amirian, S. Glüge, and T. Stadelmann, “Bias, awareness, and ignorance in deep-learning-based face recognition,” AI and Ethics, vol. 2, no. 3, pp. 509–522, Oct. 2021, doi: 10.1007/s43681-021-00108-6.
WEHRLI, Samuel, Corinna HERTWECK, Mohammadreza AMIRIAN, Stefan GLÜGE und Thilo STADELMANN, 2021. Bias, awareness, and ignorance in deep-learning-based face recognition. AI and Ethics. 27 Oktober 2021. Bd. 2, Nr. 3, S. 509–522. DOI 10.1007/s43681-021-00108-6
Wehrli, Samuel, Corinna Hertweck, Mohammadreza Amirian, Stefan Glüge, and Thilo Stadelmann. 2021. “Bias, Awareness, and Ignorance in Deep-Learning-Based Face Recognition.” AI and Ethics 2 (3): 509–22. https://doi.org/10.1007/s43681-021-00108-6.
Wehrli, Samuel, et al. “Bias, Awareness, and Ignorance in Deep-Learning-Based Face Recognition.” AI and Ethics, vol. 2, no. 3, Oct. 2021, pp. 509–22, https://doi.org/10.1007/s43681-021-00108-6.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.