Federated learning frameworks like https://github.com/adap/flower are of particular interest to healthcare research, to work seamlessly with auton-survival the following features are needed:
Easy and clean access to the model's PyTorch component
PyTorch module pre-initialisation before fit
Make the Breslow estimator optional as it contains too much info about the dataset to be safe in a federated setup
Federated learning frameworks like https://github.com/adap/flower are of particular interest to healthcare research, to work seamlessly with
auton-survivalthe following features are needed: