Federated Learning with Azure Machine Learning NVIDIA FLARE and MONAI
summary
Federated pipeline uses Azure Machine Learning, NVIDIA FLARE and MONAI framework to train multimodal 3D brain tumor segmentation on the BraTS dataset across four clinical workspaces (Europe, Canada, US, Australia). Startup YAML packages, containerized conda environments and Azure ML jobs deploy a FLARE server and clients; MLflow streams per-client metrics and logs. A simulator run precedes the distributed experiment. After 200 federated rounds the average Dice score matches centralized training but requires more rounds, showing privacy-preserving parity at higher communication cost.
AI-written summary · about
chapters (7)
- 0:00Introduction to project
- 1:20Federated learning overview
- 5:22Medical use case
- 8:12Introduction to Nvidia Flare
- 10:50Solution architecture
- 18:30Federated training demo
- 24:05Results and conclusion
related talks
30mFederated Learning from simulation to production
31mFrom Theory to Practice: Advances in Real-World Federated Learning
34mLFN AI Taskforce: Platform-Driven Fede...
26mDecentralized Federated Machine Learning: Empowering Edge Devices with Kubernetes
21mWhat, Why and How of Federated Machine Learning – Implementation Using...
31mFederated Machine Learning with Kubernetes
31mFederated Learning: So Close and Yet So Out of Reach
45mManaged federated learning on AWS: A case study for healthcare