Influence of Label and Selection Bias on Fairness Interventions

Magali Legast, Toon Calders, Francois Fouss
Proceedings of Fourth European Workshop on Algorithmic Fairness, PMLR 294:329-334, 2025.

Abstract

Bias can be introduced in different ways in machine learning datasets, with the bias type influencing the effectiveness of fairness interventions. In this work, we model fair worlds and their biased counterparts by introducing controlled label and selection bias in real-life datasets with low discrimination. We then analyze the resulting prediction models, with or without bias mitigation. Our results provide some guidance on the use of reweighing, massaging and Fairness Through Unawareness, and show that other dataset characteristics also play a role on fairness intervention efficiency, calling for further research.

Cite this Paper

BibTeX
@InProceedings{pmlr-v294-legast25a, title = {Influence of Label and Selection Bias on Fairness Interventions}, author = {Legast, Magali and Calders, Toon and Fouss, Francois}, booktitle = {Proceedings of Fourth European Workshop on Algorithmic Fairness}, pages = {329--334}, year = {2025}, editor = {Weerts, Hilde and Pechenizkiy, Mykola and Allhutter, Doris and Corrêa, Ana Maria and Grote, Thomas and Liem, Cynthia}, volume = {294}, series = {Proceedings of Machine Learning Research}, month = {30 Jun--02 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v294/main/assets/legast25a/legast25a.pdf}, url = {https://proceedings.mlr.press/v294/legast25a.html}, abstract = {Bias can be introduced in different ways in machine learning datasets, with the bias type influencing the effectiveness of fairness interventions. In this work, we model fair worlds and their biased counterparts by introducing controlled label and selection bias in real-life datasets with low discrimination. We then analyze the resulting prediction models, with or without bias mitigation. Our results provide some guidance on the use of reweighing, massaging and Fairness Through Unawareness, and show that other dataset characteristics also play a role on fairness intervention efficiency, calling for further research.} }
Endnote
%0 Conference Paper %T Influence of Label and Selection Bias on Fairness Interventions %A Magali Legast %A Toon Calders %A Francois Fouss %B Proceedings of Fourth European Workshop on Algorithmic Fairness %C Proceedings of Machine Learning Research %D 2025 %E Hilde Weerts %E Mykola Pechenizkiy %E Doris Allhutter %E Ana Maria Corrêa %E Thomas Grote %E Cynthia Liem %F pmlr-v294-legast25a %I PMLR %P 329--334 %U https://proceedings.mlr.press/v294/legast25a.html %V 294 %X Bias can be introduced in different ways in machine learning datasets, with the bias type influencing the effectiveness of fairness interventions. In this work, we model fair worlds and their biased counterparts by introducing controlled label and selection bias in real-life datasets with low discrimination. We then analyze the resulting prediction models, with or without bias mitigation. Our results provide some guidance on the use of reweighing, massaging and Fairness Through Unawareness, and show that other dataset characteristics also play a role on fairness intervention efficiency, calling for further research.
APA
Legast, M., Calders, T. & Fouss, F.. (2025). Influence of Label and Selection Bias on Fairness Interventions. Proceedings of Fourth European Workshop on Algorithmic Fairness, in Proceedings of Machine Learning Research 294:329-334 Available from https://proceedings.mlr.press/v294/legast25a.html.

Related Material