openfl_semantic_seg

Federated Learning with Openfl and Pytorch

Goal of this project

This project aims to perform a federated and distributed learning approch to semantic segmentation with Pytorch using the director pattern from Openfl

![openfl][openfl]

How to use

envoy

Simply start the envoys using their associated bash files :

    chmod +x start_envoy.sh 
    ./start_envoy.sh $ENVOY_NAME $DIRECTOR_ADDR

with parameters $ENVOY_NAME (name of the envoy, ex : env_one) $DIRECTOR_ADDR (ip address of the director, ex: localhost)

director

Start the director with

    chmod +x start_director.sh
    ./start_director.sh 

Experiment

You can then try to run the notebooks in the research folder

Run commands

Create director workspace for research

fx director create-workspace -p path/to/director_workspace_dir

Create Envoy workspace for collaborator manager

fx envoy create-workspace -p path/to/envoy_workspace_dir

Start director and disable mTLS protection

fx director start --disable-tls -c director.yaml

Start envoy and disable mTLS protection

ENVOY_NAME=orange_client_1 fx envoy start
-n "$ENVOY_NAME"
--disable-tls
--envoy-config-path envoy_config.yaml
-dh director_fqdn
-dp port

Start director with federation PKI Certificates

fx director start -c director.yaml
-rc cert/root_ca.crt
-pk cert/priv.key
-oc cert/open.crt

Start envoy with federation PKI Certificates

ENVOY_NAME=orange_client1

fx envoy start
-n "$ENVOY_NAME"
--envoy-config-path envoy_config.yaml
-dh director_fqdn
-dp port
-rc cert/root_ca.crt
-pk cert/"$ENVOY_NAME".key
-oc cert/"$ENVOY_NAME".crt