CAUSALI-T-AI
CAUSALIty Teams up with Artificial Intelligence
"Felix qui potuit rerum cognoscere causas" (Georgics, Virgil 29 BC), Fortunate, who was able to know the hidden causes of things.
More about the history of causality : Master Thesis of S. Meyahoui (in french)
The project CAUSALI-T-AI is a PEPR project funded by the ANR (2023-2028). Our consortium, gathering applied mathematicians and computer scientists, aims to tackle the causal modeling challenges using advanced machine learning approaches along four main directions :
stable causal modeling, with the objective of unraveling the difficulties resulting from the scarcity of available data and the requirement for identifiability of the learned model
learning of causal representations aiming to discover rich latent structures and construct representations that are both useful for machine learning tasks and causally well-founded.
causal inference, aiming to adapt causal models from a source domain to target domains, and design interventions to achieve a given goal
causal learning in uncertain environments
.