EUVIMEDEUVIMED entdecken
Crossref

Fairness Is Not Universal: Towards Contextualized and Autonomous Fairness in Federated Learning

Yi Zhou, Naman Goel

Proceedings of the AAAI Symposium Series · 06.11.2026 · 10 (1) · 477-486

DOI: 10.1609/aaaiss.v10i1.43442 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Symposium Series, online erschienen am 06.11.2026. Verfasst von Yi Zhou und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Much of the existing work in fairness-aware federated learning (FL) treats fairness as a global property, i.e. a constraint or objective that can be defined and applied uniformly across all clients. In this paper, we argue that this paradigm is fundamentally insufficient. Fairness is contextual, population-dependent, and often governed by domain- or jurisdiction-specific values. We evaluate fair FL methods against two client-side post-training interventions: model output post-processing and final-layer fine-tuning. Across four datasets of different modalities (tabular, signal, and image) and multiple data heterogeneity settings, we show that even when all clients nominally commit to the same fairness metric, fairness-aware FL algorithms do not provide consistent context-sensitive fairness improvements. We find that the post-training interventions are more promising, but at the same time note that, as they currently stand, they remain only partial remedies. Our empirical diagnosis helps identify the future research directions for novel FL frameworks that explicitly support contextual and autonomous fairness.

Zitieren

APA

Zhou, Y., & Goel, N. (2026). Fairness Is Not Universal: Towards Contextualized and Autonomous Fairness in Federated Learning. Proceedings of the AAAI Symposium Series, 10(1), 477-486. https://doi.org/10.1609/aaaiss.v10i1.43442

Vancouver

Zhou Y, Goel N. Fairness Is Not Universal: Towards Contextualized and Autonomous Fairness in Federated Learning. Proceedings of the AAAI Symposium Series. 2026;10(1):477-486. doi:10.1609/aaaiss.v10i1.43442

BibTeX
@article{Zhou2026,
  author = {Zhou, Yi and Goel, Naman},
  title = {Fairness Is Not Universal: Towards Contextualized and Autonomous Fairness in Federated Learning},
  journal = {Proceedings of the AAAI Symposium Series},
  year = {2026},
  volume = {10},
  number = {1},
  pages = {477--486},
  doi = {10.1609/aaaiss.v10i1.43442}
}
Kostenlos starten →