A two-stage framework that applies fairness post-processing locally on each federated learning client consistently reduces equalized odds gaps across tabular, ECG, and X-ray datasets.
Bayesian Nonparametric Federated Learning of Neural Networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
In federated learning problems, data is scattered across different servers and exchanging or pooling it is often impractical or prohibited. We develop a Bayesian nonparametric framework for federated learning with neural networks. Each data server is assumed to provide local neural network weights, which are modeled through our framework. We then develop an inference approach that allows us to synthesize a more expressive global network without additional supervision, data pooling and with as few as a single communication round. We then demonstrate the efficacy of our approach on federated learning problems simulated from two popular image classification datasets.
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
A Post-Processing-Based Fair Federated Learning Framework
A two-stage framework that applies fairness post-processing locally on each federated learning client consistently reduces equalized odds gaps across tabular, ECG, and X-ray datasets.