Guest talk: Javier Fernandez-Marques "Federated Learning Systems: taking the training to the data with Flower"

Federated Learning Systems: taking the training to the data with Flower

Javier Fernandez-Marques
Flower Labs

Abstract: This lecture will be divided into three parts: first, we will introduce the opportunities that FL brings compared to traditional centralised training; second, we will cover the main challenges, from the systems and ML perspective, that arise when deploying and running FL workloads; finally, and after an introduction of the key components of the Flower framework, we will go through three FL projects of increasing sophistication. Critically, we will see how the same FL client workload can transparently be executed in devices such as your personal computer, a Raspberry Pi, or a multi-node GPU cluster by means of Flower’s Simulation Engine. By the end of this lecture you will have all the tools needed to design and execute your own FL projects with Flower.

Bio: Javier Fernandez-Marques is a research scientist at Flower Labs. He works on the core framework and develops the Flower Simulation Engine, which allows to run Federated Learning workloads in a resource-aware manner and scale these to thousands of active clients. Javier interests lie in the intersection of Machine Learning and Systems, and more concretely running on-device ML workloads, a key component in Federated Learning. He often writes about running ML on embedded devices such as RaspberryPi or NVIDIA Jetsons. Javier got his PhD in Computer Science from the University of Oxford in 2021. Before joining Flower Labs, he worked as a research scientist at Samsung AI (Cambridge, UK)

This guest talk is hosted by Associate Professor Alex Jung, Department of Computer Science.