As a group of si-guo-jun-qi chess game players, we have a dream to create an AI agent , somehow like alphaGo, to play the Si-Guo-Jun-Qi. But for the moment, we need some sort of necessary AI knowledge to make the magic happen. From alphaGo , we guess the deep reinforecement learning may be the feasible way, therefore we decide to dive to the reinforement learning as the first step.
This repository is created just for learning the RL. The recommended resources are listed as below:
- The MOOC : Reinforcement learning from Udacity . https://cn.udacity.com/course/reinforcement-learning--ud600
- The book: Reinforcement learning: an introduction 2nd Edition. http://webdocs.cs.ualberta.ca/~sutton/book/the-book-2nd.html It is highly recommended to run the source code programmed for this book and take your own experiments.
- The lecture given by David Silver: http://www0.cs.ucl.ac.uk/staff/D.Silver/web/Teaching.html David silver now seems working in DeepMind team.
The programming lauguage is not decided yet but Python will be the first option. Today, Python is supported by most of the machine learning frameworks and becomes a de facto standard programming language in this domain. There are lots of python code bases and ML libraries that could be leveraged.
Updated: Finally, python is taken as the primary language.
Contraction mapping is one of the most important theorems applied in RL. Bellman equation is built on it.
where the source codes live