D4N Datalake for Nuclear

The initiative of the Nuclear Valley competitiveness cluster on the perimeter of Artificial Intelligence for Non-Destructive Testing has led to the creation of the D4N - Datalake for Nuclear project.

The D4N consortium has built the project by creating a secure, qualitative, reliable and anonymized data lake (DataLake) for the nuclear industry.

Making decision making by Artificial Intelligence acceptable and auditable will benefit the quality and performance of projects.

A MarketPlace has been created to register, use and market data and artificial intelligence applications on use cases, the first of which is the detection of the clogging rate of an EDF steam generator.

This project finances the PhD thesis of Gael Marcadet, on secure protocols for federated learning. The thesis is directed by Pascal Lafourcade and co-advised by Radu Ciucanu and Marta Soare.