Fed-BioMed v4.0 is now available. Here are some key new features:

  • Improved ML security: Differential Privacy (local and central) for Pytorch
  • Improved GUI security: user accounts management and authentication
  • Improved Node-side security: training plan approval mechanism
  • Support for biomedical data: MedicalFolderDataset supporting imaging + csv data
  • Integration with open biomedical datasets: IXI brain imaging, and FLamby repository
  • New tutorials: medical image segmentation, filtering datasets based on number of samples
  • Several backend improvements: mini-batch SGD for scikit-learn, validation of training arguments, redesign of training plan, etc..

More details about the new features can be found in the Fed-BioMed CHANGELOG.