https://medium.com/@monocosmo77/use-cases-of-multi-armed-bandits-part3-machine-learning-reinforcement-learning-87a35b599af6