DFL

an implementation of Federated Learning on mnist dataset using MPI.

Preparation

1. install mpi implementation in your computer
2. install python dependency

   pip3 install keras tensorflow mpi4py

Centralised Federated Learning

To run:

mpiexec -n <number of nodes> python3 centralisedFL.py

To measure the time elapsed:

time mpiexec -n <number of nodes> python3 centralisedFL.py

Vanilla MNIST Convolutional Neural Netowork

To run:

python3 original.py

To measure the time elapsed:

time python3 original.py

Experimentation

The result of the experementation can be seen in the result folder.