https://winder.ai/rl-presentation-how-to-overcome-reinforcement-learning-challenges/