Federated Learning for Intelligent Systems Across Data, Service and System Heterogeneity

About this talk

With the rapid advancement of artificial intelligence, sensing technologies, and connected computing, intelligent systems are increasingly expected to learn from data generated across everyday devices, autonomous platforms, healthcare systems, industrial environments, and cyber-physical infrastructure. However, such data often contain sensitive, private, or organization-specific information, making centralized data collection impractical or undesirable due to privacy, security, ownership, and regulatory concerns. Federated learning has emerged as a promising paradigm for enabling collaborative model training while keeping raw data local. Yet in practical intelligent systems, participating agents often differ in the data they observe, the resources they have, and the environments in which they operate. This talk examines how heterogeneity across data, devices, and systems creates fundamental challenges and motivates new algorithmic and system-level designs for federated learning. The talk further highlights how federated learning can support scalable, adaptive, and trustworthy intelligence in distributed cyber-physical systems, edge AI, and networked autonomous platforms.