EdgeLake-FL: An Automated Federated Learning Platform for the Edge
summary
EdgeLake-FL automates federated learning by using EdgeLake as the data layer and a blockchain shared metadata service to synchronize tasks so data never leaves nodes. A published LSTM training app is pulled by participating sites, three edge nodes plus an optional aggregator run five federated iterations in the demo and produce about 82% demo accuracy (integrator reports 95–96% in production). Platform is hardware-agnostic and requires only publishing the training application.
AI-written summary · about
chapters (12)
- 0:00EdgeLake-FL introduction
- 1:03Edge data management
- 3:32Federated vs conventional learning
- 5:40Why federated learning unused
- 8:12EdgeLake enabling federated learning
- 10:19Federated workflow with EdgeLake
- 16:39Hospital energy optimization
- 19:29Live demo setup
- 20:32Training and aggregation
- 24:15Model accuracy and inference
- 25:45EdgeLake-FL benefits
- 27:54Q&A hardware adoption security
related talks
48mFederated ML for Edge Applications
42mFederated Learning: Unlocking the Values of Discrete Data for Enterprise
25mData collaboration with Federated Learning
41mFederated Learning: Machine Learning on Decentralized Data
34mLFN AI Taskforce: Platform-Driven Fede...
21mWhat, Why and How of Federated Machine Learning – Implementation Using...
26mDecentralized Federated Machine Learning: Empowering Edge Devices with Kubernetes
31mFrom Theory to Practice: Advances in Real-World Federated Learning