OpenFL: Building Better AI Models with Private Data
summary
OpenFL is an open source federated learning framework that Intel and partners release to run multi‑site training without moving patient data, using gRPC, NumPy payloads, and Docker/PyPI installs. A 71‑site healthcare federation is runnable and yields measurable improvements — 71 sites across six continents and local segmentation gains of 27% and 33% (out‑of‑sample 15% and 27%). Optional hardware attestation and confidentiality use Intel SGX enclaves for trusted execution.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 0:43Federated Learning overview
- 1:48Centralized data sharing challenges
- 3:36Centralized vs Federated Learning
- 5:20Federated Learning step-by-step
- 6:03Collaborative learning techniques
- 8:12Federated Learning origins
- 8:56Introducing OpenFL
- 11:03OpenFL architecture
- 15:28OpenFL progress
- 17:11Security risks in FL
- 18:58Intel SGX and TEE
- 21:50Governor service
- 23:15Brain tumor segmentation
- 26:23Industry applications
- 27:46Contribute to OpenFL
- 28:29Q&A
related talks
26mOpenFL (Federated Learning) Building better AI models with private data
32mOpenFL: A Federated Learning Project to Power (and Secure) Your Projects
31mFrom Theory to Practice: Advances in Real-World Federated Learning
24mAI, Health & Confidentiality
18mEnabling Medical AI through Federated Learning
42mFederated Learning: Unlocking the Values of Discrete Data for Enterprise
31mFederated Learning: So Close and Yet So Out of Reach
27mFederated Learning with Azure Machine Learning NVIDIA FLARE and MONAI