Pilot 8 Partners Present Research at IEEE FLTA 2026
Partners from O-CEI Pilot 8 participated in the IEEE International Conference on Federated Learning Technologies and Applications (FLTA 2026). During the conference, researcher Mohsen Salimi Khanghah presented the paper “Dynamic Proxy-Based Clustering for Reliable Federated and Split Learning under Heterogeneity”, co-authored with Khalid Ali, Phil Aupke, Andreas Kassler, Nabil Abdennadher and Giovanna Di Marzo Serugendo.
The paper explores an adaptive approach to Federated Learning and Split Learning, using privacy-preserving proxies and information-geometric distances to dynamically adjust client grouping and aggregation topologies across training rounds. This research is relevant to O-CEI’s work on distributed and federated computing, addressing the challenge of managing heterogeneous participants while supporting reliable and privacy-preserving learning across the Cloud-Edge-IoT continuum.
