Description

Implementation of federated learning workflow. This workflow is fairly customizable with several parameters to tune. Additionally, there is native (and simple) support for MNIST and FashionMNIST for lighter workloads and CIFAR-10 and CIFAR-100 for a slightly heftier workload w.r.t. training.

Fixes #22

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Refactoring (internal implementation changes)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update (no changes to the code)
• CI change (changes to CI workflows, packages, templates, etc.)
• Version changes (changes to the package or dependency versions)

Testing

No formal testing (i.e., unit tests). But running the workflow locally can be done via:

python -m webs.run fed-learn \
  --executor process-pool --data-name mnist \
  --data-root "./"  --download true \
  --num-rounds 3 --participation 0.5 \
  --num-clients 10 --train true