Personalized Recommendation Models in Federated Settings: A Survey

TL;DR

A foundational definition of personalization in a federated setting is established, emphasizing personalized models as a critical solution for capturing fine-grained user preferences in FedRecSys.

Abstract

Federated recommender systems (FedRecSys) have emerged as a pivotal solution for privacy-aware recommendations, balancing growing demands for data security and personalized experiences. Current research efforts predominantly concentrate on adapting traditional recommendation architectures to federated environments, optimizing communication efficiency, and mitigating security vulnerabilities. However, user personalization modeling, which is essential for capturing heterogeneous preferences in this decentralized and non-IID data setting, remains underexplored. This survey addresses this gap by systematically exploring personalization in FedRecSys, charting its evolution from centralized paradigms to federated-specific innovations. We establish a foundational definition of personalization in a federated setting, emphasizing personalized models as a critical solution for capturing fine-grained user preferences. The work critically examines the technical hurdles of building personalized FedRecSys and synthesizes promising methodologies to meet these challenges. As the first consolidated study in this domain, this survey serves as both a technical reference and a catalyst for advancing personalized FedRecSys research.

Figures (8)

• Figure 1: Personalization technique comparison in centralized and federated RecSys. The colorful module denotes the user-specific parameters and the gray module represents the user-shared parameters. FL's ability to collaboratively train multiple models across different devices naturally supports the development of personalized models, making it easier to tailor recommendations to individual user needs.
• Figure 2: Overview of this paper. We summarize existing FedRecSys methods from two perspectives: RecSys Adaptation (focusing on model architectures and scenarios) and FL Enhancement (improving security, robustness, and efficiency). We then explore the role of personalization modeling in FedRecSys, emphasizing its potential for future development. Finally, we discuss challenges and solutions for personalized model-driven FedRecSys and outline promising future directions to advance research in this field.
• Figure 3: The framework of FedRecSys. The users (clients) store personal data and train the recommendation model locally. A cloud server orchestrates the global training by aggregating and distributing model parameters of all users iteratively. Once the training converges, each client device can predict the potentially interesting items for the user.
• Figure 4: Challenges (C) and solutions (S) summary for developing personalized models-driven FedRecSys.
• Figure 5: Solution schematic diagram to memory and computation overhead challenge for static item identifiers methods.