A COMPREHENSIVE SURVEY ON PERSONALIZED FEDERATED LEARNING

Abstract

Federated learning (FL) has emerged as a promising paradigm for distributed machine learning that preserves data privacy while enabling collaborative model training. However, standard FL assumes all clients share identical task objectives, which often proves unrealistic in practical applications where clients exhibit heterogeneous data distributions and varying learning goals. Personalized federated learning (PFL) addresses this fundamental limitation by enabling clients to learn personalized models tailored to their specific requirements while leveraging the benefits of distributed collaboration. This survey provides a comprehensive overview of personalized federated learning, covering: (1) fundamental concepts and motivations; (2) classification of PFL approaches; (3) key challenges; and (4) open research directions. We also discuss the trade-offs between personalization and generalization, analyze existing solutions, and identify future challenges in this rapidly evolving field.