Uncertainty Quantification in Federated Granger Causality Learning
The authors address how uncertainty propagates through Federated Granger Causality (FedGC) to provide a principled way to identify reliable cross-client interactions in vertically partitioned data. They derive closed-form covariance recursions and show that steady-state uncertainty depends only on client data statistics (aleatoric), not model parameter priors, leading to a robust post-training hypothesis testing method.