tartarini/dafne-server

Backend server to store and distributed federated machine learning models.

Dafne-Server

This is the backend server for the Dafne platform.

The server does the following:

1. provide trained models to the client
2. receive updated models from the client
3. merge the new client models with the existing model to improve the model (federated learning)
4. provide new improved model to the client

How to start the server

Put models into the following folder structure:

db
'-> model
    '-> Classifier
        '-> XXX.model
    '-> Leg
        '-> XXX.model
    '-> Thigh
        '-> XXX.model

Replace XXX with an integer (unique ID / timestamp) (e.g. 1603281013.model).
Then start the server:

python main.py

How to use the API

See test_api.py

Build and use docker container

Build:

docker build --no-cache -t dafne-server:master .

Run:

docker run -p 5000:80 -v /mnt/data/dafne-server-db:/app/db dafne-server:master

Managing cloud server

see details