Fair Clustering: A Causal Perspective

Fritz Bayer, Drago Plečko, Niko Beerenwinkel, Jack Kuipers
Proceedings of the Fourth Conference on Causal Learning and Reasoning, PMLR 275:340-358, 2025.

Abstract

Clustering algorithms may unintentionally propagate or intensify existing disparities, leading to unfair representations or biased decision-making. Current fair clustering methods rely on notions of fairness that do not capture any information on the underlying causal mechanisms. We show that optimising for non-causal fairness notions can paradoxically induce direct discriminatory effects from a causal standpoint. We present a clustering approach that incorporates causal fairness metrics to provide a more nuanced approach to fairness in unsupervised learning. Our approach enables the specification of the causal fairness metrics that should be minimised. We demonstrate the efficacy of our methodology using datasets known to harbour unfair biases.

Cite this Paper

BibTeX
@InProceedings{pmlr-v275-bayer25a, title = {Fair Clustering: A Causal Perspective}, author = {Bayer, Fritz and Ple\v{c}ko, Drago and Beerenwinkel, Niko and Kuipers, Jack}, booktitle = {Proceedings of the Fourth Conference on Causal Learning and Reasoning}, pages = {340--358}, year = {2025}, editor = {Huang, Biwei and Drton, Mathias}, volume = {275}, series = {Proceedings of Machine Learning Research}, month = {07--09 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v275/main/assets/bayer25a/bayer25a.pdf}, url = {https://proceedings.mlr.press/v275/bayer25a.html}, abstract = {Clustering algorithms may unintentionally propagate or intensify existing disparities, leading to unfair representations or biased decision-making. Current fair clustering methods rely on notions of fairness that do not capture any information on the underlying causal mechanisms. We show that optimising for non-causal fairness notions can paradoxically induce direct discriminatory effects from a causal standpoint. We present a clustering approach that incorporates causal fairness metrics to provide a more nuanced approach to fairness in unsupervised learning. Our approach enables the specification of the causal fairness metrics that should be minimised. We demonstrate the efficacy of our methodology using datasets known to harbour unfair biases.} }
Endnote
%0 Conference Paper %T Fair Clustering: A Causal Perspective %A Fritz Bayer %A Drago Plečko %A Niko Beerenwinkel %A Jack Kuipers %B Proceedings of the Fourth Conference on Causal Learning and Reasoning %C Proceedings of Machine Learning Research %D 2025 %E Biwei Huang %E Mathias Drton %F pmlr-v275-bayer25a %I PMLR %P 340--358 %U https://proceedings.mlr.press/v275/bayer25a.html %V 275 %X Clustering algorithms may unintentionally propagate or intensify existing disparities, leading to unfair representations or biased decision-making. Current fair clustering methods rely on notions of fairness that do not capture any information on the underlying causal mechanisms. We show that optimising for non-causal fairness notions can paradoxically induce direct discriminatory effects from a causal standpoint. We present a clustering approach that incorporates causal fairness metrics to provide a more nuanced approach to fairness in unsupervised learning. Our approach enables the specification of the causal fairness metrics that should be minimised. We demonstrate the efficacy of our methodology using datasets known to harbour unfair biases.
APA
Bayer, F., Plečko, D., Beerenwinkel, N. & Kuipers, J.. (2025). Fair Clustering: A Causal Perspective. Proceedings of the Fourth Conference on Causal Learning and Reasoning, in Proceedings of Machine Learning Research 275:340-358 Available from https://proceedings.mlr.press/v275/bayer25a.html.

Related Material