Title: Federated Learning in Food Safety Surveillance: Architectures, Applications, and Open Challenges
Authors: Akinode John Lekan
Opeoluwa Oluyemisi Oyinlola
Volume: 10
Issue: 6
Pages: 20-25
Publication Date: 2026/06/28
Abstract:
The world food supply chains are increasingly being challenged to ensure safety and quality. As the chains break up and become geographically dispersed, traditional centralized surveillance systems fall behind. They rely on data processing in pools, which leads to conflicts with data sovereignty, business confidentiality, and privacy laws.This review explores Federated Learning, a decentralised solution that allows multiple stakeholders to collaboratively learn a model without sharing raw data. The study looks at the advancements that have been published in recent years and assesses the potential of edge-enabled and blockchain-backed federated systems in enhancing food safety monitoring A systematic review of the new federated architectures and their application in food safety was carried out. Potential application scenarios for pathogen detection, spoilage prediction, fraud detection and traceability along the supply chain are explored. Various horizontal and vertical data partitioning options were evaluated to determine their suitability for different scenarios of operations.Decentralised federated frameworks always performed better than isolated local models. Their ability to balance privacy and performance was fairly similar to in the case of fully centralised systems. Blockchain-integrated and edge-deployed configurations proved to be especially promising for real-time cold-chain and contamination monitoring.Federated Learning is a promising approach to achieving privacy-preserving and high-performance food safety intelligence. However, there are still challenges that need to be addressed with respect to non-independent and identically distributed (non-IID) data and communication overhead in resource-constrained environments. To protect public health in a digitalised global food system, the path is going towards zero-trust federated frameworks