技术控

    今日:51| 主题:49312
收藏本版 (1)
最新软件应用技术尽在掌握

[其他] Learning Reinforcement Learning (with Code, Exercises and Solutions)

[复制链接]
ζ扯蛋的夏天 发表于 2016-10-2 22:18:31
97 0

立即注册CoLaBug.com会员,免费获得投稿人的专业资料,享用更多功能,玩转个人品牌!

您需要 登录 才可以下载或查看,没有帐号?立即注册

x
Skip all the talk and go directly to the Github Repo with code and exercises.         
    Why Study Reinforcement Learning

    Reinforcement Learning is one of the fields I’m most excited about. Over the past few years amazing results like      learning to play Atari Games from raw pixelsand      Mastering the Game of Gohave gotten a lot of attention, but RL is also widely used in Robotics, Image Processing and Natural Language Processing.   
    Combining Reinforcement Learning and Deep Learning techniques works extremely well. Both fields heavily influence each other. On the Reinforcement Learning side Deep Neural Networks are used as function approximators to learn good representations, e.g. to process Atari game images or to understand the board state of Go. In the other direction, RL techniques are making their way into supervised problems usually tackled by Deep Learning. For example, RL techniques are used to implement attention mechanisms in image processing, or to optimize long-term rewards in conversational interfaces and neural translation systems. Finally, as Reinforcement Learning is concerned with making optimal decisions it has some extremely interesting parallels to human Psychology and Neuroscience (and many other fields).
    With lots of open problems and opportunities for fundamental research I think we’ll be seeing multiple Reinforcement Learning breakthroughs in the coming years. And what could be more fun than teaching machines to play Starcraft and Doom?
    How to Study Reinforcement Learning

    There are many excellent Reinforcement Learning resources out there. Two I recommend the most are:
   
          
  •         David Silver’s Reinforcement Learning Course      
  •         Richard Sutton’s & Andrew Barto’s Reinforcement Learning: An Introduction (2nd Edition)book.   
    The latter is still work in progress but it’s ~80% complete. The course is based on the book so the two work quite well together. In fact, these two cover almost everything you need to know to understand most of the recent research papers. The prerequisites are basic Math and some knowledge of Machine Learning.
    That covers the theory. But what about practical resources? What about actually implementing the algorithms that are covered in the book/course? That’s where this post and the      Github repositorycomes in. I’ve tried to implement most of the standard Reinforcement Algorithms using Python,      OpenAI Gymand Tensorflow. I separated them into chapters (with brief summaries) and exercises and solutions so that you can use them to supplement the theoretical material above.      All of this is in the Github repository.   
    Some of the more time-intensive algorithms are still work in progress, so feel free to contribute. I’ll update this post as I implement them.
    Table of Contents

   
          
  •         Introduction to RL problems, OpenAI gym      
  •         MDPs and Bellman Equations      
  •         Dynamic Programming: Model-Based RL, Policy Iteration and Value Iteration      
  •         Monte Carlo Model-Free Prediction & Control      
  •         Temporal Difference Model-Free Prediction & Control      
  •         Function Approximation      
  •         Deep Q Learning(WIP)      
  •         Policy Gradient Methods(WIP)      
  • Learning and Planning (WIP)      
  • Exploration and Exploitation (WIP)   
    List of Implemented Algorithms

   
          
  •                   Dynamic Programming Policy Evaluation        
          
  •                   Dynamic Programming Policy Iteration        
          
  •         Dynamic Programming Value Iteration      
  •         Monte Carlo Prediction      
  •         Monte Carlo Control with Epsilon-Greedy Policies      
  •         Monte Carlo Off-Policy Control with Importance Sampling      
  •         SARSA (On Policy TD Learning)      
  •         Q-Learning (Off Policy TD Learning)      
  •         Q-Learning with Linear Function Approximation      
  •         Deep Q-Learning for Atari Games      
  •         Double Deep-Q Learning for Atari Games      
  • Deep Q-Learning with Prioritized Experience Replay (WIP)      
  •         Policy Gradient: REINFORCE with Baseline      
  •         Policy Gradient: Actor Critic with Baseline      
  •         Policy Gradient: Actor Critic with Baseline for Continuous Action Spaces      
  • Deterministic Policy Gradients for Continuous Action Spaces (WIP)      
  • Deep Deterministic Policy Gradients (DDPG) (WIP)      
  • Asynchronous Advantage Actor Critic (A3C) (WIP)   
友荐云推荐




上一篇:Linux日常使用总结
下一篇:Story of Equality in .Net - Part 6
酷辣虫提示酷辣虫禁止发表任何与中华人民共和国法律有抵触的内容!所有内容由用户发布,并不代表酷辣虫的观点,酷辣虫无法对用户发布内容真实性提供任何的保证,请自行验证并承担风险与后果。如您有版权、违规等问题,请通过"联系我们"或"违规举报"告知我们处理。

*滑动验证:
您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

我要投稿

推荐阅读

扫码访问 @iTTTTT瑞翔 的微博
回页顶回复上一篇下一篇回列表手机版
手机版/CoLaBug.com ( 粤ICP备05003221号 | 文网文[2010]257号 )|网站地图 酷辣虫

© 2001-2016 Comsenz Inc. Design: Dean. DiscuzFans.

返回顶部 返回列表