An example of using NVIDIA FLARE to train a medical image analysis model using federated averaging (FedAvg) and a MONAI Bundle.
This example will also guide you on using MONAI FL with FLARE to collect client data statistics and visualize both global and local intensity histograms using FLARE's FL simulator.
The example can be extended to use additional FL algorithms available in NVIDIA FLARE.
FL deployment requires secure provisioning and an admin API to submit jobs. This example runs you through the process and includes instructions on running FedAvg with experiment tracking using MLflow and homomorphic encryption for secure server-side aggregation.
In this example, we use an already prepared provisioning file (project.yml) to run experiments on a single machine. For real-world deployment, additional considerations must be taken into account. Please see the real-world FL docs for further details on using FL in a real-world deployment.