Skip to content

Latest commit

 

History

History
52 lines (34 loc) · 1.37 KB

README.md

File metadata and controls

52 lines (34 loc) · 1.37 KB
Raw

Serverless Federated Learning

The assignment component within the academic course 'VU Serverless Computing WS-2023.'

Technologies used

Purpose Technology
Serverless platform Openwhisk
Object Storage Minio
Coordination / Messaging Kafka
Machine Learning scikit-learn
Local Development Environment kind ~ Kubernetes in Docker

TODOs / Issues

  • currently the actions need to be manually warmed up before starting training.
  • EMNIST instead of MNIST
  • add links to the above technologies

Setting up the resources necessary for FL

Run:

./setup-everything.sh

Optionally change permissions so it is executable:

chomd +x setup-everything.sh

Starting FL

After setup the main-app, it gets served on localhost:5000.

To start training navigate to /learn and optionally pass the arguments with url params: /learn?nclients=10&nrounds=10&nselected=5

Default values are:

  • nclients=3
  • nselected=3
  • nrounds=5

Each round starts nclients and each client gets one split. As soon as the 50 splits get used the splits start getting re-used. After nselected model updates have been received, the round is finished and these are merged and saved as a new global model, which is then used in the next round.

Visualization

The solution is visualized in the following diagram:

FL