Description 📜

This demo demonstrates a federated learning using Flwr framework with a private Etherum blockchain to classify images in CIFAR-10 dataset.

The project contains the following:

The code related to smart contracts is placed under ./Blockchain

The code related to server organisation is placed under ./Server

The code related to client organisation is placed under ./Client

Requirement ⚙️

To run this project successfully, you need to install the following:

Anacoda

Set up enviroment by using venv.yml with Anacoda promp conda env create --file venv.yml

NodeJS

Truffle suite and Ganache UI to simulate the blockchain

How to use 🧑‍🏫

Setup blockchain

In ./Blockchain, compile the smart contracts by truffle compile.
After compilation, run the promp truffle migrate --network development to mirgrate the contract on the network.
Please help check the configuration in truffle-config.js
Ganache should be opened before running the project

For IPFS

Change your keys in api_key.json

For Server organisation

1. Change the direction to ./Server
2. Use uvicorn promp to start a server: uvicorn server_api:server --reload. The server will run default on localhost:8000
3. Access localhost:8000/docs to interact with the API

For Client organisation

1. Change the direction to ./Client
2. Use uvicorn promp to start a client: uvicorn client_api:app --reload --port PORT_NUMBER. The client will run default on localhost:PORT_NUMBER
3. Access localhost:PORT_NUMBER/docs to interact with the API

Demo 📹

Demo Video