Modern Reinforcement Learning Techniques to Deal with Large Action Spaces