Episode 12: Federated Learning
16 November, 2021
Data privacy is one of the most important challenges facing artificial intelligence. To build models, it is often necessary to have access to all the data, and in the case of particularly sensitive data, it involves numerous ethical issues. Federated learning was born to offer a viable alternative: building large-scale models using many small models, each trained on small data sets. In this way, each small model only uses the data it has access to, and the large model only has access to the parameters of these trained models, and never to their data. In this episode we talk about federated learning with Nuria Rodriguez Barroso, PhD student at the Andalusian Institute of Artificial Intelligence – DaSCI-.
Some of you may remember her from our first episode about women in computer science, returns to SintonIA. Nuria is a PhD student at DaSCI and participates in the recent research project entitled “Federated Learning for Preserving Data Privacy”. Moreover, her specialty is precisely on the topic of adversarial attacks in federated learning.