世界模型（World Model）实验以及原理

Basic Concepts in Reinforcement Learning

世界模型的实验

[World models on colab]{https://colab.research.google.com,/drive/1sF2iUdhMbm2mwdECvy01l9iWEFiIodlb#scrollTo=pCg_b9DOwDN6}

补充：在colab上显示游戏环境

通过xvfb,我们可以很方便的在colab上观察训练过程与模型训练结果。

demo: Policy Gradient display on Colab

Reference

• On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models

• Guest Post (Part I): Demystifying Deep Reinforcement Learning

• Simple reinforcement learning methods to learn CartPole
• One Big Net For Everything
• Simple reinforcement learning methods to learn CartPole
• Hallucinogenic Deep Reinforcement Learning Using Python and Keras