p16 / Preprint
Federated Learning in Real-World Applications: Mitigating the Data Availability Problem
Author-posted position paper · 2025
In plain language
A position paper on a tension in federated learning: keeping records decentralized can conflict with applications that need continuous access to original data.
Research context
Privacy-conscious deployment must also plan for the availability of information needed by clinicians or other authorized users.
Approach
Healthcare motivates discussion of encrypted backups, decentralized storage, replication, hierarchical edge approaches, and distributed cryptographic protocols.
Findings
The paper identifies data availability as an adoption barrier and outlines possible mitigation strategies rather than reporting a comparative benchmark.
Summary based on full paper. Author-posted manuscript dated January 2025. A formal publication venue was not verified.