Preserving Privacy in Next Keyword Prediction using Federated Learning
Shruti Tyagi, Aditya Pawar, Ujjwal Kumar, Dhairya Ameria · 2023
Federated learning has emerged as a ground-breaking approach for collaborative model training across decentralized devices, addressing the challenges of centralized data storage and privacy concerns. In the context of recommendation systems, federated learning enables the creation of personalized recommendations while upholding user privacy. This paper introduces the "FedScope Recommendation System" (FRS) designed to predict next keywords. FRS leverages FedScope architecture which enhances the flexibility and enables the development of a tailored recommendation system. By adopting the modular aspects of FedScope, FRS crafts a recommendation system aligned with the client-server cluster's unique requirements. The proposed FRS serves as an innovative illustration of how federated learning can revolutionize personalized recommendations while maintaining privacy and security.