Scalable federated learning: privacy, efficiency, and personalization at scale
Abstract
Federated Learning (FL) has emerged as a transformative paradigm for training machine learning models across decentralized data sources while preserving data locality. By enabling collaborative learning without centralizing raw data, FL addresses critical concerns related to privacy, regulatory compliance, and data sovereignty. However, deploying FL at real-world scale introduces fundamental challenges arising from statistical heterogeneity, system variability, communication constraints, and evolving privacy threats. This chapter provides a comprehensive and structured survey of algorithmic advances that enable scalable federated learning, organized around a central trilemma involving privacy preservation, communication and computational efficiency, and effective personalization. We systematically analyze optimization strategies, privacy-enhancing mechanisms, and personalization frameworks, emphasizing their interdependencies and inherent trade-offs. The chapter further examines frontier applications in healthcare, wireless edge intelligence, and federated adaptation of large language models, and discusses system-level implementations. Finally, we outline open research challenges and future directions toward building robust, scalable, and trustworthy federated learning ecosystems.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.