CNCF-hosted Co-located Events North America 2024

Embracing machine learning on edge devices brings unprecedented challenges, particularly in enhancing models over time while safeguarding data privacy. Enter Federated Machine Learning, a paradigm enabling model training across multiple edge devices or servers without data exchange. This talk elucidates the fundamentals of ML on decentralized data and highlights the disparities in conventional approaches. Leveraging Kubernetes, we demonstrate how to orchestrate Federated Machine Learning at scale, facilitating ML model training and diverse computations. A live demo showcases image classification, showcasing lower latency, reduced power consumption, and enhanced privacy—all achievable at scale with Kubernetes. Join us to revolutionize model development with privacy-preserving, Kubernetes-powered Federated Machine Learning.