After a brief presentation of Reinforcement Learning in general, I give the theoretical bases for a particular reinforcement algorithm, Q-Learning, and its neural-network-powered version, Deep Q-Learning. Then I show some difficulties that can be encountered during implementation and I suggest solutions to overcome them, drawn from my personal experience.
Reference: the slides and videos from David Silver’s online course on Reinforcement Learning (http://www0.cs.ucl.ac.uk/staff/d.silver/web/Teaching.html).