Ensembles in machine learning: theory and practice

Ensembles combine the predictions of several machine learning models, and are used ubiquitously in machine learning. We will present examples of such ensembles, focusing in particular on random forests, deep ensembles, and dropout ensembles. We will also discuss the theory of ensembles, and in particular under which conditions collective predictions are better than individual ones. Along the way, we will encounter connections with several fields related to collective intelligence (e.g. political science or philosophy).