The Research Lab of Wendelin Böhmer at the TU Delft. We investigate theoretical questions in deep reinforcement learing, in particular deduction, adaptation and out-of-distribution generalization. Böhmer/Boehmer/Bohmer Lab