aiqbench analysis
FairMean addresses the critical tension between promoting fairness in distributed learning and maintaining robustness against label poisoning attacks by introducing bounded gradient weighting. This approach effectively mitigates the amplification of poisoned gradients while improving accuracy for underperforming clients without compromising overall model stability. The research offers a significant advancement in secure federated learning architectures where data privacy and equitable performance are paramount.
FairMean: Promoting Fairness in Distributed Learning under Label Poisoning Attacks
arXiv:2609.26377v1 Announce Type: cross Abstract: Fairness-aware distributed learning prioritizes clients with large losses to reduce performance disparities, but label poisoning can create large losses, thereby inducing a fairness--robustness conflict. We propose FairMean to manage this conflict. FairMean weights client gradients using a bounded, nondecreasing function of local loss. The increasing weights prioritize high-loss clients to promote fairness, while the upper bound prevents excessive loss-induced amplification of poisoned-client gradients. In the absence of label poisoning, we show that minimizing the FairMean objective is more conducive to solution fairness than minimizing the standard average-loss objective. Under label poisoning, we establish an average-stationarity bound whose attack-dependent term is proportional to the square of the poisoned-client fraction. Experiments show that FairMean promotes fairness by reducing accuracy variance while improving worst-client accuracy.
Abstract:Fairness-aware distributed learning prioritizes clients with large losses to reduce performance disparities, but label poisoning can create large losses, thereby inducing a fairness--robustness conflict. We propose FairMean to manage this conflict. FairMean weights client gradients using a bounded, nondecreasing function of local loss. The increasing weights prioritize high-loss clients to promote fairness, while the upper bound prevents excessive loss-induced amplification of poisoned-client gradients. In the absence of label poisoning, we show that minimizing the FairMean objective is more conducive to solution fairness than minimizing the standard average-loss objective. Under label poisoning, we establish an average-stationarity bound whose attack-dependent term is proportional to the square of the poisoned-client fraction. Experiments show that FairMean promotes fairness by reducing accuracy variance while improving worst-client accuracy.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2609.26377 [cs.LG]
(or arXiv:2609.26377v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.26377 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 219 words
Related coverage
[AINews] The Future of Latent Space
Latent Space · Sep 25, 2026
J++ Lens: Jacobian Filtering Enables More Faithful Workspace Lenses
LessWrong · Oct 8, 2026
A Framework for Evaluating and Benchmarking Concept Drift Detection Methods
arXiv · Jun 9, 2026
The Most Forbidden Technique is not always forbidden
LessWrong · Jul 18, 2026