We were delighted to see the Open Data Institute selected the ground breaking work led by Andrew Soltan as a case study in their most recent PETs explainer. Over 130k patient records were safely processed during this study with Flower acting as the federated learning framework adopted. Check out the explainer today! Mohammad Naseri built a wonderful companion hands-on tutorial using Flower that fits alongside the ODI introduction to PETs!
The Open Data Institute have developed an interactive explainer about Privacy Enhancing Technologies, like federated learning. I'm delighted that our paper in The Lancet Digital Health was chosen as a case study for the explainer. We aimed to make it easy for hospitals to contribute to developing & testing AI models, using micro-computers to reduce the technical burden on hospitals. We piloted the platform across four UK hospital groups to build and evaluate a screening test for COVID-19, while custody of the data remained with the hospital, and showed the approach improved performance. Many thanks to Flower Labs (Daniel J. Beutel & Nicholas Lane) for supporting the online demo page - our work was built upon the friendly and easy-to-implement Flower library. Thanks also to Claudine Tinsman and team for involving me in the ODI's report, and Anshul Thakur, Prof Tingting Zhu, Prof David Eyre, & Prof David Clifton! Please see the interactive explainer here: https://lnkd.in/eqACSedB Report: https://lnkd.in/eYjR3bkd https://lnkd.in/e8Zja52g Link to our paper: https://lnkd.in/eVucSsJf