How to Adopt Emerging Federated Learning Technologies
summary
Flame claims to let teams swap and extend federated learning topologies without rewriting ML code by using a Tag topology graph of roles and channels that the control plane expands into workers, aggregators and networking. It runs on Kubernetes (including minikube and a single-node flame-in-a-box), supports P2P and MQTT backends, integrates MLflow for model registry, and shows experimentally that hierarchical and hybrid topologies change time-to-accuracy and personalization tradeoffs versus vanilla FedAvg.
AI-written summary · about
chapters (13)
- 0:00Federated learning intro
- 1:04What is federated learning
- 4:59Why federated learning needed
- 6:23Adopting FL challenges
- 7:28Conventional FL limitations
- 9:12Introducing Flame framework
- 11:19Flame topology design
- 16:04Flame SDK and base classes
- 19:23Flame demo overview
- 20:25Building jobs in Flame
- 23:57Flame future & thanks
- 24:58Flame adoption benefits
- 25:47Q&A on FL use cases
related talks
21mWhat, Why and How of Federated Machine Learning – Implementation Using...
48mFederated ML for Edge Applications
31mFrom Theory to Practice: Advances in Real-World Federated Learning
42mFederated Learning: Unlocking the Values of Discrete Data for Enterprise
30mFederated Learning from simulation to production
04mProject Lightning Talk: Empowering Federated Learning with Multi-Cluster Management for Privacy and Efficiency
57mKeynote: Federated Learning with Flower: AI's Next Frontier
32mInnovation in the Age of Regulation: Federated Learning with Flower