https://analyticsindiamag.com/deepmind-found-new-approach-to-create-faster-reinforcement-learning-models/