Biomathematics Seminar: causal feature selection.

A Biomathematics Seminar held at Imperial College London, open to all on a first come, first served basis. The speaker is Emilie Devijver of Université Grenoble Alpes, whose talk is titled 'Causal feature selection for out-of-distribution generalization'. She discusses how causal inference, developed to move beyond simple correlations, can support explainability, generalization and action, and proposes a causal feature selection for out-of-distribution prediction with latent confounders, using causal invariance to make predictions in a new environment. The work is in collaboration with PhD student Théotime Le Goff and is hosted by Barbara Bravi.