A Privacy-Preserving Federated Collaborative Filtering Framework for Secure Recommender Systems

This paper aims to enhance FL performance while protecting privacy by reducing noise during anonymization, using feature selection for dimensionality reduction, and generalizing data. Developing a predictive model for healthcare applications is the goal of this study, which also examines feature-based data separation rather than record-based data separation and assesses the suggested model's effectiveness using actual healthcare data. A recommender system is typically tailored to provide useful and effective suggestions for a specific type of item, such as CDs or news. Its design, graphical user interface, and core recommendation technique are all adjusted to cater to this specific type of item. Product recommendations, movie and TV program suggestions, article suggestions, and countless more examples are among the most common and significant use cases.