Federated Machine Learning with Flower: Packaging, Monitoring, Securing

Some practical details, summarised:

• The event officially starts at 18:00 at the Thoughtworks office, Prinsengracht 741-4 and will last till 21:00
• Food and pizza will be provided during the event
• If you have any dietary restrictions, please let us know as soon as possible
• At 5:30 we will have a coding environment setup session for those who are more comfortable with setting it up together
• The repo for the event can be found here: https://github.com/mlops-and-crafts/federated-learning please have a look!
• Minimum requirements: Python 3.9, docker, good vibes and willingness to learn :)

Requirements: Python 3.9, docker, good vibes and willingness to learn :)

In this Meetup we will discover how a federated learning model works. Over the course of the workshop we will deploy a federated learning model on the edge together. Sara and Artiom will be guiding you on how to package, monitor and secure this model. We will begin with a brief overview of the fundamentals of federated learning, then dive into the three key topics in-depth.

The deployment portion of the workshop will focus on packaging, training the models and deployment. Each participant will represent an edge in the federated model. You will be training the model on your device, and we will observe together as the live server federates the different edge models.

In the monitoring section of the workshop we will discuss different tools and techniques available for monitoring federated learning models in production. We will talk about the importance of monitoring model performance, data drift and other key metrics.

What happens when a malicious actor intercepts federated model updating? The security and privacy portion of the workshop will cover some techniques and strategies available for protecting federated learning models.

By the end of the workshop, you will have a better understanding of how to deploy, monitor and secure federated learning models. Food and drink during the event is on us! Afterwards, there will be ample opportunity to stick around and network at the Thoughtworks office. Join us for this first edition of MLOps and Crafts and let's learn together!

Meetup hosts:

Sara Perricone

Sara is a Data Scientist at Thoughtworks with a research background in bioinformatics and over four years experience carrying out various data science projects throughout different industries. Sara has actively contributed to the development, deployment and maintenance of machine learning models, within these she has worked on a federated learning project where user privacy is of great importance. Her areas of expertise include data exploration & analysis, machine learning modelling and experiment design.

Artiom Troyanovsky

Artiom works as a Machine Learning Engineer at Thoughtworks. His main expertise lies in Bayesian methods. He is passionate about productionization of data science and machine learning solutions.

GitHub Repo:
https://github.com/mlops-and-crafts/federated-learning/