Federated Learning (FL) offers a privacy-preserving machine learning method by enabling collaborative model training across multiple clients without data sharing, securing sensitive information at its source. This talk explores Machine Learning applications and how to keep them secure, for example in critical sectors like healthcare.
Collaborative learning, and in particular Federated Learning (FL) is a Machine Learning approach in which multiple clients collaboratively train a Neural Network model on their private data without the need to share the data. This strategy guarantees that data stays in its initial location, never being disclosed to external entities.
This talk, will cover an introduction on how FL is used and its advantages when supporting secured data collaboration projects for example in environments like health care, where it is not possible to publish patient data for Machine Learning purposes. We will focus on the security perspective of Machine Learning and privacy attacks and defenses in those systems.
This introduction draws upon the teachings of a course conducted by Phillip Rieger from System Security Lab at TU Darmstadt.