# Learning World Graphs to Accelerate Hierarchical Reinforcement Learning ### Key Ideas from the paper * The approach ensures that the worker only has to care about going from one hop to the next hop, making it very easy for the worker. The manager only has a constrained set of "important states" available, this makes its *search space* much more condensed. * The challenge is gathering the world graph 😅 * The real question is how do you learn the world graph? --- Date: 20211112 Links to: [Reinforcement Learning (old)](Reinforcement%20Learning%20(old).md) Tags: References: * [Paper on Arxiv](https://arxiv.org/abs/1907.00664) * [Youtube paper review](https://youtu.be/Qk4lJdp7ZAs?t=541)