Hochschule Karlsruhe Hochschule Karlsruhe - University of Applied Sciences
NewsCalendarStaffEmergency_crisis
Hochschule Karlsruhe Hochschule Karlsruhe - University of Applied Sciences Hochschule Karlsruhe Hochschule Karlsruhe - University of Applied Sciences
Hochschule Karlsruhe Hochschule Karlsruhe - University of Applied Sciences

Publication on Analyzing the Impact of Participant Failures in Cross-Silo Federated Learning at FLTA 2025

In this work, we analyze the influence of missing participants in Cross-Silo Federated Learning with heterogeneous data and investigate how different aspects modify the impact. We show that a missing participant can lead to an optimistic evaluation and the timing influences the quality of the resulting model.

Participant reliability is a critical challenge in cross-silo federated learning, where organizations collaboratively train models without sharing their data. When participants fail to contribute during training, it can significantly affect model quality and evaluation. We are excited to announce that our paper “Analyzing the Impact of Participant Failures in Cross-Silo Federated Learning”, authored by Fabian Stricker, David Bermbach and Christian Zirpins, has been published in the Proceedings of the International Conference on Federated Learning Technologies and Applications (FLTA) 2025 – for details please use this link.

This work was co-funded by the German Federal Ministry of Education and Research (BMBF) under Grant 13FH587KX1 (FederatedForecasts). If you’d like to learn more about the paper, you can read it here.  

To explore the research project in more detail, please visit our website.