README.md

Python Secure Decentralized Learning App

A Blockchain application with federated learning model using neural networks.

What is blockchain? How it is implemented? And how it works?

Please read the step-by-step implementation tutorial to get your answers :)

What are neural networks? And how it works?

Please read the step-by-step tutorial to get your answers :)

Instructions to run

Clone the project,

$ git clone https://github.com/Mandeep3838/Blockchain-machinelearning.git

Install the dependencies,

$ cd Blockchain-machinelearning
$ pip install -r requirements.txt

Start a blockchain-federated learning model,

$ chmod 777 miner_script.sh clients_script.sh kill_sessions.sh
$ ./miner_script.sh

On a different Terminal

$ ./clients_script.sh

After this step

• Two instances of blockchain node are now up and running on port 8000,8001.
• Client nodes are running on ports {5000,5001,...,5010}
• Testing node is running on port 9000
• You can check results on the firefox browser
• Error files for each client and test node are saved in the working directory as blocks are added.
• For checking chain, refresh http://127.0.0.1:8000/chain

Working

1. Demo Video

2. Miner Node

3. Client Node

Different Scenarios

• Default app takes two client nodes as malicious, you can increase the maliciousness by changing z variable in the clients_script.sh
• Increasing Network (Requires Higher Computational power)
  • You can increase the miner and client nodes by increasing the limit of for loops in miner_script.sh, clients_script.sh
  • In Miner Script
    • Replace 8001 by 80xx
  • In Client Script
    • More curl lines to set up a network.
    • Add new peer in the peers array.
    • Replace 5010 by 50yy, 8001 by 80xx
    • Modify start and end variable, as the share of data available to each client from 1100 rows.
  • In kill_session Script
    • Replace 5010 by 50yy, 8001 by 80xx
  • Run the bash script

Documentation of Code

Present in the docs directory