This repository provides an implementation of several reinforcement learning algorithms, must read papers/books and questions (of any kind).

In addition, I keep this repository updated with my thoughts and future work (mainly to answer the question: How to improve reinforcement learning for non-stationary environments).
The long term goal of this repository is to be the must-have of GitHub Reinforcement Learning repository (RL Algo, environements, papers, etc).

A first report (available in french and soon in english) proposed an in-dept study of some of the algorithms used. Note that the algorithms are not written because they are easily found on the internet. For the first version of the report, my work was based on this document : Reinforcement Learning.

non-exhaustive list of what's coming soon:

• Algorithms:
  1. Deep Q-Network
  2. PPO
  3. Actor-Critic
• Environment:
  1. My own environments

All the articles I've read and plan to read. I keep track of the teams involved so that I can get an idea of the research themes of RL teams around the world.

• Complexity of Planning with Partial Observability
  Teams involved : Albert-Ludwigs-Universität Freiburg, Institut für Informatik
• An introduction to Reinforcement Learning and its video
  Researchers involved : Richard S. Sutton and Andrew G. Barto
• World Models
  Teams involved : Google Brain, NNAISENSE and Swiss AI Lab, IDSIA
• Gans and its analysis
  Teams involved : Université de Montreal
• Learn more about (finite) MDPs
  Researchers involved : Richard S. Sutton and Andrew G. Barto
• Outracing champion Gran Turismo drivers with deep reinforcement learning
  Teams involved : Sony AI
• Hierarchical Reinforcement Learning for Precise Soccer Shooting Skills using a Quadrupedal Robot
  Teams involved : RAIL lab and MILA
• Improving Intrinsic Exploration with Language Abstractions
  Teams involved : Stanford NLP Group, Stanford AI Lab, Allen School's Natural Language Processing META AI Researh, DARK Lab and Cohere
• Exploration via Elliptical Episodic Bonuses and OpenReview
  Teams involved : META AI Researh and DARK Lab
• Accelerated Quality-Diversity through Massive Parallelism
  Teams involved : Adaptive & Intelligent Robotics Lab at the Imperial College London
• Discovering and Exploiting Sparse Rewards in a Learned Behavior Space
  Teams involved : AI Lab, SoftBank Robotics Europe and Institut des Systémes Intelligents et de Robotique, ISIR
• Sparse Reward Exploration via Novelty Search and Emitters
  Teams involved : AI Lab, SoftBank Robotics Europe and Institut des Systémes Intelligents et de Robotique, ISIR
• Emergence of Spatial Coordinates via Exploration
  Team involved : AI Lab, SoftBank Robotics Europe
• Generalization in Cooperative Multi-Agent Systems
  Team involved : WhiRL and DARK Lab
• (MuZero) Mastering Atari, Go, chess and shogi by planning with a learned model📝
  Team involved : DeepMind
• (AlphaZero) Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm📝
  Team involved : DeepMind
• (AlphaGo Zero) Mastering the game of Go without human knowledge📝
  Team involved : DeepMind
• (AlphaGO) Mastering the game of GO with deep neural networks and tree search📝
  Team involved : DeepMind
• (AlphaFold) Mastering the game of GO with deep neural networks and tree search📝
  Team involved : DeepMind
• (AlphaTensor) Discovering faster matrix multiplication algorithms with Reinforcement Learning📝
  Team involved : DeepMind
• (AlphaFold) Highly accurate protein structure prediction with AlphaFold📝
  Team involved : DeepMind
• (DeepNash) Mastering the Game of Stratego with Model-Free Multiagent Reinforcement Learning📝
  Team involved : DeepMind
• ETA Prediction with Graph Neural Networks in Google Maps📝
  Team involved : DeepMind, Google
• Reward is enough📝
  Team involved : Google

• Writing a research article: advice to beginners📝
  
• How to do Research At the MIT AI Lab
  
• Writing a Good Research Paper by Vincent Lepetit
  
• How to write a great research paper: Slides and Video by Simon Peyton Jones
  
• How to give a great research talk: Slides, Video and Paper
  

• Making the World Differentiable
• Player of Games
• Approximate exploitability: Learning a best response in large games
• Towards a Better Understanding of Representation Dynamics under TD-learning
• MAESTRO: OPEN-ENDED ENVIRONMENT DESIGN FOR MULTI-AGENT REINFORCEMENT LEARNING
• Deep reinforcement learning with double q-learning
• The road to modern AI

Here are the courses I took to further my knowledge of Reinforcement Learning:

• Reinforcement Learning - MVA by Alessandro Lazaric
  
• Reinforcement Learning - Scool by Philippe Preux
  
• Game Theory and Applications by Bruno Tuffin and Patrick Maillé
  
• Markov Chains by Bruno Tuffin and Bruno Sericola
  
• Deep Learning with Python by François Chollet
• Deep Learning for Computer Vision by Justin Johnson
  

Here are some books that I read or plan to read:

• Mathematics of Statistical Sequential Decision Making by Odalric-Ambrym Maillard
  
• Statistical Learning and Sequential Prediction by Karthik Sridharan and Sasha Rakhlin
• Algorithms for decision making by Mykel J. Kochenderfer, Tim A. Wheeler and Kyle H. Wray
• Multi-Agent Reinforcement Learning: Foundations and Modern Approaches by Stefano V. Albrecht, Filippos Christianos and Lukas Schäfer
  
• An introduction to Reinforcement Learning by Richard S. Sutton and Andrew G. Barto
  
• Software Engineering at Google and the book in short SWE at Google in short

Here are some blogs, videos or webpages that I found interseting:

• CONFERENCE JENSEN HUANG (NVIDIA) and ILYA SUTSKEVER (OPEN AI): AI TODAY AND VISION OF THE FUTURE📝 by Ilya Sutskever and Jensen Huang
  
• Quality-Diversity optimisation algorithms by Antoine Cully, Jean-Baptiste Mouret and Stephane Doncieux
  
• I am Jürgen Schmidhuber, Ask Me Anything! by Jürgen Schmidhuber