Enabling Medical AI through Federated Learning
summary
Medical AI needs diverse training data but hospitals rarely share raw records so federated evaluation first sends a global model to sites for local metrics and then federated learning aggregates weight updates instead of moving data, using techniques like homomorphic encryption and differential privacy. The team builds and open sources the Flower framework and demonstrates it in research and industry settings including 50–60 hospitals and partners like Harvard, showing it can reduce bias in tasks such as skin cancer detection by enabling broad validation across sites via federated evaluation.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 0:54About the speaker
- 1:47Why do we need medical AI
- 2:47How to do a machine learning project
- 8:18Very learning
- 11:17Questions
related talks
32mInnovation in the Age of Regulation: Federated Learning with Flower
26mMachine Learning without the Data #federatedlearning
57mKeynote: Federated Learning with Flower: AI's Next Frontier
31mFrom Theory to Practice: Advances in Real-World Federated Learning
32mPlant a Touch-Me-Not: Train Models Without Anyone Touching Your Data with Flower
45mManaged federated learning on AWS: A case study for healthcare
27mTalk by Micah Sheller (Flower/ML Commons)
28mA Privacy Enhancing Federated Learning Framework for Multi-Party Distributed ..