https://www.reddit.com/r/reinforcementlearning/comments/1ta3bnz/currently_experimenting_with_exploration_policies/