Proximity-Based Self-Federated Learning
Date:
I presented our paper “Proximity-Based Self-Federated Learning” in the main track of ACSOS 2024.
The talk introduced a decentralized federated learning approach in which devices autonomously form multiple learning federations according to their spatial proximity and local model behaviour. The proposed method aims to improve personalization and robustness under non-IID data while reducing dependence on a single central coordinator.
ACSOS offered an ideal venue for discussing the work from the perspective of self-organization, autonomic computing, and collective adaptive systems.