Self-Paced Context Evaluation for Contextual Reinforcement Learning

           Theresa Eimer 1 Andre Biedenkapp 2 Frank Hutter 2 3 Marius Lindauer 1

             Abstract
   Reinforcement learning (RL) has made a lot of
   advances for solving a single problem in a given
   environment; but learning policies that generalize
   to unseen variations of a problem remains chal-        Figure 1: Example instances of the contextual PointMass
   lenging. To improve sample efficiency for learn-       ...