Abstract Federated learning is a machine learning technique in which the model is trained across a number of decentralized devices (clients) without the need to move the data to a central location. Because users’ data is stored on their devices rather than a central server or third party, this method gives consumers improved privacy and security. Contrarily, centralized learning mandates that users submit their data to a central server, raising issues with data security and privacy. The purpose of the current study was to evaluate and compare the effectiveness of the centralized and federated learning paradigms in the context of a simple regression task using simulated data. The findings demonstrated that while protecting user privacy, federated learning may attain accuracy levels that are on par with those of centralized learning. Our study also demonstrated the viability of implementing federated learning using well-known machine learning frameworks like TensorFlow Federated.
Alan : Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|