Training AI algorithms today requires more and more data, and that data is increasingly complicated to find. This is even more evident in the digital security and identity sectors, where AI is becoming increasingly important to keep citizens safe around the world.
The problem is that data contains personal information about individuals, so the goal is to find new ways to train algorithms while respecting the privacy of individuals and ensuring that there is no bias. This also requires respecting the legislation on the data used, such as the European GDPR.
This challenge aims to find new ways to create databases for model development, whether by anonymizing information, creating synthetic data or using frugal learning.
Projects will be related to biometric authentication or identification algorithms.